PM / QoS: Introduce latency tolerance device PM QoS type
Subsystems (11)
cover PM / QoS: Introduce latency tolerance device PM QoS type
Hi All,
On some platforms hardware may switch to an energy-saving mode on the fly on the basis of certain utilization metrics used by it. That usually is desirable from the energy conservation standpoint, but it generally causes latencies to increase which may adversely affect some operations. For this reason, the platforms in question usually provide some interfaces for software to indicate its latency tolerance and possibly to prevent the energy-saving modes from being selected too aggressively.
The following series of patches introduces a device PM QoS type allowing those interfaces to be used by kernel code and user space. It is designed in analogy with the existing resume latency device PM QoS type, which allows some pieces of the existing device PM QoS code to be re-used and makes the new user space interface fit into the existing framework.
Patch [1/5] modifies the names of symbols, variables, functions and structure fields associated with the existing resume latency device PM QoS type to avoid any confusion with the new one introduced by the subsequent patches.
Patch [2/5] introduces a new field in struct pm_qos_constraints for specifying a special value to be returned as the effective requirement when the given list of PM QoS requirements is empty. That field is necessary for the new latency tolerance device PM QoS type.
Patch [3/5] introduces the latency tolerance device PM QoS type along with documentation.
Patch [4/5] modifies the ACPI LPSS (Low-Power Subsystem) driver to hook up LPSS devices to the new latency tolerance device PM QoS interface.
Patch [5/5] modifies the dev_pm_qos_add_ancestor_request() routine so that it can be used by drivers of devices without hardware latency tolerance support for specifying their requirements via the ancestors of those devies.
Comments welcome!
Thanks,
Rafael
Signature · 2 lines
I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
PATCH 1/5 PM / QoS: Rename device resume latency QoS items
From: Rafael J. Wysocki <[email protected]>
Rename symbols, variables, functions and structure fields related do the resume latency device PM QoS type so that it is clear where they belong (in particular, to avoid confusion with the latency tolerance device PM QoS type introduced by a subsequent changeset).
Update the PM QoS documentation to better reflect its current state.
--- Documentation/power/pm_qos_interface.txt | 37 +++++++++----------- Documentation/trace/events-power.txt | 2 - drivers/base/power/power.h | 4 +- drivers/base/power/qos.c | 55 +++++++++++++++---------------- drivers/base/power/sysfs.c | 32 +++++++++--------- drivers/mtd/nand/sh_flctl.c | 2 - include/linux/pm_qos.h | 14 +++---- include/trace/events/power.h | 4 +- 8 files changed, 73 insertions(+), 77 deletions(-)
Documentation/trace/events-power.txt +1 -1
@@ -92,5 +92,5 @@ dev_pm_qos_remove_request "devi The first parameter gives the device name which tries to add/update/remove QoS requests. -The second parameter gives the request type (e.g. "DEV_PM_QOS_LATENCY"). +The second parameter gives the request type (e.g. "DEV_PM_QOS_RESUME_LATENCY"). The third parameter is value to be added/updated/removed.
drivers/base/power/qos.c +27 -28
151 lines not on this page · raw message
drivers/mtd/nand/sh_flctl.c +1 -1
@@ -897,7 +897,7 @@ static void flctl_select_chip(struct mtdif (!flctl->qos_request) { ret =dev_pm_qos_add_request (&flctl->pdev->dev, &flctl->pm_qos, -DEV_PM_QOS_LATENCY , +DEV_PM_QOS_RESUME_LATENCY ,100 );if (ret <0 ) dev_err(&flctl->pdev->dev,
include/linux/pm_qos.h +7 -7
@@ -32,7 +32,7 @@ enum pm_qos_flags_status {#define PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0 -#define PM_QOS_DEV_LAT_DEFAULT_VALUE 0 +#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -49,7 +49,7 @@ struct pm_qos_flags_request { };enum dev_pm_qos_req_type { -DEV_PM_QOS_LATENCY =1 , +DEV_PM_QOS_RESUME_LATENCY =1 ,DEV_PM_QOS_FLAGS , };
@@ -87,9 +87,9 @@ struct pm_qos_flags { };struct dev_pm_qos { -struct pm_qos_constraints latency; +struct pm_qos_constraints resume_latency;struct pm_qos_flags flags; -struct dev_pm_qos_request *latency_req; +struct dev_pm_qos_request *resume_latency_req;struct dev_pm_qos_request *flags_req; };
@@ -196,9 +196,9 @@ int dev_pm_qos_expose_flags(struct devicvoid dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); -static inline s32 dev_pm_qos_requested_latency (struct device *dev) +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) { -return dev->power.qos->latency_req->data.pnode.prio; +return dev->power.qos->resume_latency_req->data.pnode.prio; }static inline s32 dev_pm_qos_requested_flags (struct device *dev)
@@ -215,7 +215,7 @@ static inline void dev_pm_qos_hide_flagsstatic inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } -static inline s32 dev_pm_qos_requested_latency (struct device *dev) {return 0 ; } +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }#endif
include/trace/events/power.h +2 -2
@@ -359,8 +359,8 @@ DECLARE_EVENT_CLASS(dev_pm_qos_request,TP_printk ("device=%s type=%s new_value=%d" ,__get_str (name),__print_symbolic (__entry->type, - { DEV_PM_QOS_LATENCY, "DEV_PM_QOS_LATENCY" }, - { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), + { DEV_PM_QOS_RESUME_LATENCY, "DEV_PM_QOS_RESUME_LATENCY" }, + { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), __entry->new_value) );
drivers/base/power/power.h +2 -2
@@ -89,8 +89,8 @@ extern void dpm_sysfs_remove(struct deviextern void rpm_sysfs_remove (struct device *dev);extern int wakeup_sysfs_add (struct device *dev);extern void wakeup_sysfs_remove (struct device *dev); -extern int pm_qos_sysfs_add_latency (struct device *dev); -extern void pm_qos_sysfs_remove_latency (struct device *dev); +extern int pm_qos_sysfs_add_resume_latency (struct device *dev); +extern void pm_qos_sysfs_remove_resume_latency (struct device *dev);extern int pm_qos_sysfs_add_flags (struct device *dev);extern void pm_qos_sysfs_remove_flags (struct device *dev);
drivers/base/power/sysfs.c +17 -15
@@ -218,15 +218,16 @@ static ssize_t autosuspend_delay_ms_stor static DEVICE_ATTR(autosuspend_delay_ms, 0644, autosuspend_delay_ms_show, autosuspend_delay_ms_store); -static ssize_t pm_qos_latency_show (struct device *dev, -struct device_attribute *attr,char *buf) +static ssize_t pm_qos_resume_latency_show (struct device *dev, +struct device_attribute *attr, +char *buf) { -return sprintf (buf,"%d\n" ,dev_pm_qos_requested_latency (dev)); +return sprintf (buf,"%d\n" ,dev_pm_qos_requested_resume_latency (dev)); } -static ssize_t pm_qos_latency_store (struct device *dev, -struct device_attribute *attr, -const char *buf,size_t n) +static ssize_t pm_qos_resume_latency_store (struct device *dev, +struct device_attribute *attr, +const char *buf,size_t n) {s32 value;int ret;
@@ -237,12 +238,13 @@ static ssize_t pm_qos_latency_store(struif (value <0 )return -EINVAL ; - ret =dev_pm_qos_update_request (dev->power.qos->latency_req, value); + ret =dev_pm_qos_update_request (dev->power.qos->resume_latency_req, + value);return ret <0 ? ret : n; } static DEVICE_ATTR(pm_qos_resume_latency_us, 0644, - pm_qos_latency_show, pm_qos_latency_store); + pm_qos_resume_latency_show, pm_qos_resume_latency_store); static ssize_t pm_qos_no_power_off_show(struct device *dev, struct device_attribute *attr,
@@ -618,15 +620,15 @@ static struct attribute_group pm_runtime .attrs = runtime_attrs, }; -static struct attribute *pm_qos_latency_attrs[] = { +static struct attribute *pm_qos_resume_latency_attrs[] = { #ifdef CONFIG_PM_RUNTIME &dev_attr_pm_qos_resume_latency_us.attr, #endif /* CONFIG_PM_RUNTIME */NULL , }; -static struct attribute_group pm_qos_latency_attr_group = { +static struct attribute_group pm_qos_resume_latency_attr_group = { .name = power_group_name, - .attrs = pm_qos_latency_attrs, + .attrs = pm_qos_resume_latency_attrs, };static struct attribute *pm_qos_flags_attrs[] = {
@@ -681,14 +683,14 @@ void wakeup_sysfs_remove(struct device *sysfs_unmerge_group (&dev->kobj, &pm_wakeup_attr_group); } -int pm_qos_sysfs_add_latency (struct device *dev) +int pm_qos_sysfs_add_resume_latency (struct device *dev) { -return sysfs_merge_group (&dev->kobj, &pm_qos_latency_attr_group); +return sysfs_merge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); } -void pm_qos_sysfs_remove_latency (struct device *dev) +void pm_qos_sysfs_remove_resume_latency (struct device *dev) { -sysfs_unmerge_group (&dev->kobj, &pm_qos_latency_attr_group); +sysfs_unmerge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); }int pm_qos_sysfs_add_flags (struct device *dev)
Documentation/power/pm_qos_interface.txt +16 -21
@@ -89,16 +89,16 @@ node. 2. PM QoS per-device latency and flags framework For each device, there are two lists of PM QoS requests. One is maintained -along with the aggregated target of latency value and the other is for PM QoS -flags. Values are updated in response to changes of the request list. +along with the aggregated target of resume latency value and the other is for +PM QoS flags. Values are updated in response to changes of the request list. -Target latency value is simply the minimum of the request values held in the -parameter list elements. The PM QoS flags aggregate value is a gather (bitwise -OR) of all list elements' values. Two device PM QoS flags are defined currently: -PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. +Target resume latency value is simply the minimum of the request values held in +the parameter list elements. The PM QoS flags aggregate value is a gather +(bitwise OR) of all list elements' values. Two device PM QoS flags are defined +currently: PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. -Note: the aggregated target value is implemented as an atomic variable so that -reading the aggregated value does not require any locking mechanism. +Note: the aggregated target value is implemented in such a way that reading the +aggregated value does not require any locking mechanism. From kernel mode the use of this interface is the following:
@@ -137,14 +137,14 @@ Add a PM QoS request for the first direc power.ignore_children flag is unset. int dev_pm_qos_expose_latency_limit(device, value) -Add a request to the device's PM QoS list of latency constraints and create -a sysfs attribute pm_qos_resume_latency_us under the device's power directory -allowing user space to manipulate that request. +Add a request to the device's PM QoS list of resume latency constraints and +create a sysfs attribute pm_qos_resume_latency_us under the device's power +directory allowing user space to manipulate that request. void dev_pm_qos_hide_latency_limit(device) Drop the request added by dev_pm_qos_expose_latency_limit() from the device's -PM QoS list of latency constraints and remove sysfs attribute pm_qos_resume_latency_us -from the device's power directory. +PM QoS list of resume latency constraints and remove sysfs attribute +pm_qos_resume_latency_us from the device's power directory. int dev_pm_qos_expose_flags(device, value) Add a request to the device's PM QoS list of flags and create sysfs attributes
@@ -163,7 +163,7 @@ a per-device notification tree and a glo int dev_pm_qos_add_notifier(device, notifier): Adds a notification callback function for the device. The callback is called when the aggregated value of the device constraints list -is changed. +is changed (for resume latency device PM QoS only). int dev_pm_qos_remove_notifier(device, notifier): Removes the notification callback function for the device.
@@ -171,14 +171,9 @@ Removes the notification callback functi int dev_pm_qos_add_global_notifier(notifier): Adds a notification callback function in the global notification tree of the framework. -The callback is called when the aggregated value for any device is changed. +The callback is called when the aggregated value for any device is changed +(for resume latency device PM QoS only). int dev_pm_qos_remove_global_notifier(notifier): Removes the notification callback function from the global notification tree of the framework. - - -From user mode: -No API for user space access to the per-device latency constraints is provided -yet - still under discussion. -
PATCH 2/5 PM / QoS: Add no_constraints_value field to struct pm_qos_constraints
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new field, no_constraints_value, to struct pm_qos_constraints representing a list of PM QoS constraint requests to be returned by pm_qos_get_value() when that list of requests is empty.
That field will be equal to default_value for all of the existing global PM QoS classes and for the resume latency device PM QoS type, but it will be different from default_value for the new latency tolerance device PM QoS type introduced by the next changeset.
--- drivers/base/power/qos.c | 1 + include/linux/pm_qos.h | 1 + kernel/power/qos.c | 5 ++++- 3 files changed, 6 insertions(+), 1 deletion(-)
drivers/base/power/qos.c +1 -0
@@ -190,6 +190,7 @@ static int dev_pm_qos_constraints_allocaplist_head_init (&c->list); c->target_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->default_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; + c->no_constraint_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->type =PM_QOS_MIN ; c->notifiers = n;
include/linux/pm_qos.h +1 -0
@@ -77,6 +77,7 @@ struct pm_qos_constraints {struct plist_head list;s32 target_value;/* Do not change to 64 bit */ s32 default_value; +s32 no_constraint_value;enum pm_qos_type type;struct blocking_notifier_head *notifiers; };
kernel/power/qos.c +4 -1
@@ -66,6 +66,7 @@ static struct pm_qos_constraints cpu_dma .list = PLIST_HEAD_INIT(cpu_dma_constraints.list), .target_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .default_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &cpu_dma_lat_notifier, };
@@ -79,6 +80,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_lat_constraints.list), .target_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &network_lat_notifier, };
@@ -93,6 +95,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_tput_constraints.list), .target_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .type = PM_QOS_MAX, .notifiers = &network_throughput_notifier, };
@@ -128,7 +131,7 @@ static const struct file_operations pm_qstatic inline int pm_qos_get_value (struct pm_qos_constraints *c) {if (plist_head_empty (&c->list)) -return c->default_value; +return c->no_constraint_value;switch (c->type) {case PM_QOS_MIN :
PATCH 3/5 PM / QoS: Introduce latency tolerance device PM QoS type
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new latency tolerance device PM QoS type to be use for specifying active state (RPM_ACTIVE) memory access (DMA) latency tolerance requirements for devices. It may be used to prevent hardware from choosing overly aggressive energy-saving operation modes (causing too much latency to appear) for the whole platform.
This feature reqiures hardware support, so it only will be available for devices having a new .set_latency_tolerance() callback in struct dev_pm_info populated, in which case the routine pointed to by it should implement whatever is necessary to transfer the effective requirement value to the hardware.
Whenever the effective latency tolerance changes for the device, its .set_latency_tolerance() callback will be executed and the effective value will be passed to it. If that value is negative, which means that the list of latency tolerance requirements for the device is empty, the callback is expected to switch the underlying hardware latency tolerance control mechanism to an autonomous mode if available. If that value is 0, in turn, and the hardware supports a special "no requirement" setting, the callback is expected to use it. That allows software to prevent the hardware from automatically updating the device's latency tolerance in response to its power state changes (e.g. during transitions from D3cold to D0), which generally may be done in the autonomous latency tolerance control mode.
If .set_latency_tolerance() is present for the device, a new pm_qos_latency_tolerance_us attribute will be present in the devivce's power directory in sysfs. Then, user space can use that attribute to specify its latency tolerance requirement for the device, if any. Writing 0 to it means "no requirement, but do not let the hardware control latency tolerance" and using a negative value allows the hardware to be switched to the autonomous mode if there are no other requirements from the kernel side in the device's list.
--- Documentation/ABI/testing/sysfs-devices-power | 25 ++++++ Documentation/power/pm_qos_interface.txt | 61 +++++++++++++--- drivers/base/power/qos.c | 95 +++++++++++++++++++++++++- drivers/base/power/sysfs.c | 56 +++++++++++++-- include/linux/pm.h | 1 include/linux/pm_qos.h | 11 +++ kernel/power/qos.c | 13 ++- 7 files changed, 238 insertions(+), 24 deletions(-)
include/linux/pm_qos.h +11 -0
@@ -33,6 +33,8 @@ enum pm_qos_flags_status {#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT (-1)#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -50,6 +52,7 @@ struct pm_qos_flags_request {enum dev_pm_qos_req_type {DEV_PM_QOS_RESUME_LATENCY =1 , +DEV_PM_QOS_LATENCY_TOLERANCE ,DEV_PM_QOS_FLAGS , };
@@ -89,8 +92,10 @@ struct pm_qos_flags {struct dev_pm_qos {struct pm_qos_constraints resume_latency; +struct pm_qos_constraints latency_tolerance;struct pm_qos_flags flags;struct dev_pm_qos_request *resume_latency_req; +struct dev_pm_qos_request *latency_tolerance_req;struct dev_pm_qos_request *flags_req; };
@@ -196,6 +201,8 @@ void dev_pm_qos_hide_latency_limit(strucint dev_pm_qos_expose_flags (struct device *dev,s32 value);void dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); +s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev); +int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val);static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {
@@ -215,6 +222,10 @@ static inline int dev_pm_qos_expose_flagstatic inline void dev_pm_qos_hide_flags (struct device *dev) {}static inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } +static inline s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev) + {return PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT ; } +static inline int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val) + {return 0 ; }static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }
drivers/base/power/qos.c +94 -1
143 lines not on this page · raw message
drivers/base/power/sysfs.c +49 -7
85 lines not on this page · raw message
include/linux/pm.h +1 -0
@@ -583,6 +583,7 @@ struct dev_pm_info {unsigned long accounting_timestamp;#endif struct pm_subsys_data *subsys_data;/* Owned by the subsystem. */ +void (*set_latency_tolerance)(struct device *,s32 );struct dev_pm_qos *qos; };
kernel/power/qos.c +8 -5
@@ -173,6 +173,7 @@ int pm_qos_update_target(struct pm_qos_c {unsigned long flags;int prev_value, curr_value, new_value; +int ret;spin_lock_irqsave (&pm_qos_lock, flags); prev_value =pm_qos_get_value (c);
@@ -208,13 +209,15 @@ int pm_qos_update_target(struct pm_qos_ctrace_pm_qos_update_target (action, prev_value, curr_value);if (prev_value != curr_value) { -blocking_notifier_call_chain (c->notifiers, - (unsigned long )curr_value, -NULL ); -return 1 ; + ret =1 ; +if (c->notifiers) +blocking_notifier_call_chain (c->notifiers, + (unsigned long )curr_value, +NULL ); }else { -return 0 ; + ret =0 ; } +return ret; }/**
Documentation/power/pm_qos_interface.txt +51 -10
71 lines not on this page · raw message
Documentation/ABI/testing/sysfs-devices-power +24 -1
@@ -187,7 +187,7 @@ Description: Not all drivers support this attribute. If it isn't supported, attempts to read or write it will yield I/O errors. -What: /sys/devices/.../power/pm_qos_latency_us +What: /sys/devices/.../power/pm_qos_resume_latency_us Date: March 2012 Contact: Rafael J. Wysocki <[email protected]> Description:
@@ -205,6 +205,29 @@ Description: This attribute has no effect on system-wide suspend/resume and hibernation. +What: /sys/devices/.../power/pm_qos_latency_tolerance_us +Date: January 2014 +Contact: Rafael J. Wysocki <[email protected]> +Description: + The /sys/devices/.../power/pm_qos_latency_tolerance_us attribute + contains the PM QoS active state latency tolerance limit for the + given device, which is the maximum memory access latency the + device can suffer without any visible adverse effects on user + space functionality. If that value is 0, the latency does not + matter to user space at all. + + Reading "n/a" from this file means that the maximum memory + access latency for the device may be determined automatically + by the hardware as needed. Writing a negative value to it + allows the hardware to be switched to this mode if there are no + other latency tolerance requirements from the kernel side. + + This attribute is only present if the feature controlled by it + is supported by the hardware. + + This attribute has no effect on runtime suspend and resume of + devices and on system-wide suspend/resume and hibernation. + What: /sys/devices/.../power/pm_qos_no_power_off Date: September 2012 Contact: Rafael J. Wysocki <[email protected]>
PATCH 4/5 ACPI / LPSS: Support for device latency tolerance PM QoS
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Modify acpi_lpss_platform_notify() to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
--- drivers/acpi/acpi_lpss.c | 66 ++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 62 insertions(+), 4 deletions(-)
drivers/acpi/acpi_lpss.c +62 -4
@@ -33,6 +33,12 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -316,6 +322,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -337,7 +354,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -390,6 +407,46 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct acpi_device *adev =ACPI_COMPANION (dev); +struct lpss_private_data *pdata; +u32 ltr_mode, ltr_val; + +if (WARN_ON (!adev)) +return ; + + pdata =acpi_driver_data (adev); +if (WARN_ON (!pdata || !pdata->mmio_base)) +return ; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; +if (val >LPSS_LTR_MAX_VAL ) + val =LPSS_LTR_MAX_VAL ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US ; +if (val >0 ) + ltr_val |=LPSS_LTR_SNOOP_REQ ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -415,11 +472,12 @@ static int acpi_lpss_platform_notify(strreturn 0 ; } -if (action ==BUS_NOTIFY_ADD_DEVICE ) +if (action ==BUS_NOTIFY_ADD_DEVICE ) { ret =sysfs_create_group (&pdev->dev.kobj, &lpss_attr_group); -else if (action ==BUS_NOTIFY_DEL_DEVICE ) + pdev->dev.power.set_latency_tolerance = acpi_lpss_set_ltr; + }else if (action ==BUS_NOTIFY_DEL_DEVICE ) {sysfs_remove_group (&pdev->dev.kobj, &lpss_attr_group); - + }return ret; }
Re: [RFC/RFT][PATCH 4/5] ACPI / LPSS: Support for device latency tolerance PM QoS
drivers/acpi/acpi_lpss.c static int acpi_lpss_platform_notify(str · line 477
Quoted 9 lines · Rafael J. Wysocki
drivers/acpi/acpi_lpss.c +2
-2@@ -415,11 +472,12 @@ static int acpi_lpss_platform_notify(strreturn 0 ; } -if (action ==BUS_NOTIFY_ADD_DEVICE ) +if (action ==BUS_NOTIFY_ADD_DEVICE ) { ret =sysfs_create_group (&pdev->dev.kobj, &lpss_attr_group); -else if (action ==BUS_NOTIFY_DEL_DEVICE ) + pdev->dev.power.set_latency_tolerance = acpi_lpss_set_ltr;
While trying to test this I noticed that BUS_NOTIFY_ADD_DEVICE happens after call to dpm_sysfs_add(), so LTR field is never exposed to the userspace.
Quoted 7 lines
drivers/acpi/acpi_lpss.c +2
-1+ }else if (action ==BUS_NOTIFY_DEL_DEVICE ) {sysfs_remove_group (&pdev->dev.kobj, &lpss_attr_group); - + }return ret; }
Re: [RFC/RFT][PATCH 4/5] ACPI / LPSS: Support for device latency tolerance PM QoS
Quoted 5 lines · Mika Westerberg
Quoted 4 lines · Rafael J. Wysocki
drivers/acpi/acpi_lpss.c +0
-0@@ -415,11 +472,12 @@ static int acpi_lpss_platform_notify(strreturn 0 ; }
Quoted 9 lines
drivers/acpi/acpi_lpss.c static int acpi_lpss_platform_notify(str · line 477
Quoted 5 lines
drivers/acpi/acpi_lpss.c +2
-2-if (action ==BUS_NOTIFY_ADD_DEVICE ) +if (action ==BUS_NOTIFY_ADD_DEVICE ) { ret =sysfs_create_group (&pdev->dev.kobj, &lpss_attr_group); -else if (action ==BUS_NOTIFY_DEL_DEVICE ) + pdev->dev.power.set_latency_tolerance = acpi_lpss_set_ltr;While trying to test this I noticed that
BUS_NOTIFY_ADD_DEVICEhappens after call todpm_sysfs_add(), so LTR field is never exposed to the userspace.
Ahh, I confused things, thanks for reporting this!
I'll need to hook it up to acpi_platform_notify() somehow I think.
Quoted 7 lines
Quoted 7 lines
drivers/acpi/acpi_lpss.c +2
-1+ }else if (action ==BUS_NOTIFY_DEL_DEVICE ) {sysfs_remove_group (&pdev->dev.kobj, &lpss_attr_group); - + }return ret; }
Signature · 2 lines
I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
Re: [RFC/RFT][PATCH 4/5] ACPI / LPSS: Support for device latency tolerance PM QoS
Quoted 16 lines · Rafael J. Wysocki
Quoted 5 lines · Mika Westerberg
Quoted 4 lines · Rafael J. Wysocki
drivers/acpi/acpi_lpss.c +0
-0@@ -415,11 +472,12 @@ static int acpi_lpss_platform_notify(strreturn 0 ; }Quoted 9 lines
drivers/acpi/acpi_lpss.c static int acpi_lpss_platform_notify(str · line 477
Quoted 5 lines
drivers/acpi/acpi_lpss.c +2
-2-if (action ==BUS_NOTIFY_ADD_DEVICE ) +if (action ==BUS_NOTIFY_ADD_DEVICE ) { ret =sysfs_create_group (&pdev->dev.kobj, &lpss_attr_group); -else if (action ==BUS_NOTIFY_DEL_DEVICE ) + pdev->dev.power.set_latency_tolerance = acpi_lpss_set_ltr;While trying to test this I noticed that
BUS_NOTIFY_ADD_DEVICEhappens after call todpm_sysfs_add(), so LTR field is never exposed to the userspace.
Quoted 3 lines
Ahh, I confused things, thanks for reporting this!
I'll need to hook it up to
acpi_platform_notify()somehow I think.
OK, two patches to replace this one ([4/5]) will follow.
On top of them we can move the code from acpi_lpss_platform_notify() to acpi_lpss_(bind|unbind)() and get rid of acpi_lpss_nb, which I think would be an improvement too. :-)
Thanks,
Rafael
PATCH 1/1 ACPI / scan: Add bind/unbind callbacks to struct acpi_scan_handler
From: Rafael J. Wysocki <[email protected]>
In some cases it may be necessary to perform certain setup/cleanup operations on a device object representing a physical device after it has been associated with an ACPI companion by acpi_bind_one() or before disassociating it from that companion by acpi_unbind_one(), respectively. If there is a struct acpi_bus_type object for the given device's bus type, the .setup()/.cleanup() callbacks from there are executed for these purposes. However, an analogous mechanism will be necessary for devices whose bus types don't have corresponding struct acpi_bus_type objects and that have specific ACPI scan handlers.
For those devices, add new .bind() and .unbind() callbacks to struct acpi_scan_handler that will be executed by acpi_platform_notify() right after the given device has been associated with an ACPI comapnion and by acpi_platform_notify_remove() right before calling acpi_unbind_one() for that device, respectively.
--- drivers/acpi/glue.c | 12 ++++++++++++ include/acpi/acpi_bus.h | 2 ++ 2 files changed, 14 insertions(+)
drivers/acpi/glue.c +12 -0
@@ -287,6 +287,7 @@ EXPORT_SYMBOL_GPL(acpi_unbind_one);static int acpi_platform_notify (struct device *dev) {struct acpi_bus_type *type =acpi_get_bus_type (dev); +struct acpi_device *adev;int ret; ret =acpi_bind_one (dev,NULL );
@@ -303,9 +304,14 @@ static int acpi_platform_notify(struct dif (ret) goto out; } + adev =ACPI_COMPANION (dev); +if (!adev) + goto out;if (type && type->setup) type->setup (dev); +else if (adev->handler && adev->handler->bind) + adev->handler->bind (dev); out: #if ACPI_GLUE_DEBUG
@@ -324,11 +330,17 @@ static int acpi_platform_notify(struct dstatic int acpi_platform_notify_remove (struct device *dev) { +struct acpi_device *adev =ACPI_COMPANION (dev);struct acpi_bus_type *type; +if (!adev) +return 0 ; + type =acpi_get_bus_type (dev);if (type && type->cleanup) type->cleanup (dev); +else if (adev->handler && adev->handler->unbind) + adev->handler->unbind (dev);acpi_unbind_one (dev);return 0 ;
include/acpi/acpi_bus.h +2 -0
@@ -133,6 +133,8 @@ struct acpi_scan_handler {struct list_head list_node;int (*attach)(struct acpi_device *dev,const struct acpi_device_id *id);void (*detach)(struct acpi_device *dev); +void (*bind)(struct device *phys_dev); +void (*unbind)(struct device *phys_dev);struct acpi_hotplug_profile hotplug; };
PATCH 1/1 ACPI / LPSS: Support for device latency tolerance PM QoS
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Add .bind()/.unbind() callbacks to lpss_handler to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition and to clear those pointers on device removal, respectively.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
--- drivers/acpi/acpi_lpss.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 70 insertions(+), 1 deletion(-)
drivers/acpi/acpi_lpss.c +70 -1
@@ -33,6 +33,12 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -316,6 +322,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -337,7 +354,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -390,6 +407,38 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); +u32 ltr_mode, ltr_val; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; +if (val >LPSS_LTR_MAX_VAL ) + val =LPSS_LTR_MAX_VAL ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US ; +if (val >0 ) + ltr_val |=LPSS_LTR_SNOOP_REQ ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -427,9 +476,29 @@ static struct notifier_block acpi_lpss_n .notifier_call = acpi_lpss_platform_notify, }; +static void acpi_lpss_bind (struct device *dev) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); + +if (WARN_ON (!pdata || !pdata->mmio_base)) +return ; + +if (pdata->mmio_size >= pdata->dev_desc->prv_offset +LPSS_LTR_SIZE ) + dev->power.set_latency_tolerance = acpi_lpss_set_ltr; +else +dev_err (dev,"MMIO size insufficient to access LTR\n" ); +} + +static void acpi_lpss_unbind (struct device *dev) +{ + dev->power.set_latency_tolerance =NULL ; +} +static struct acpi_scan_handler lpss_handler = { .ids = acpi_lpss_device_ids, .attach = acpi_lpss_create_device, + .bind = acpi_lpss_bind, + .unbind = acpi_lpss_unbind, };void __init acpi_lpss_init (void )
PATCH 5/5 PM / QoS: Add type to dev_pm_qos_add_ancestor_request() arguments
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Rework dev_pm_qos_add_ancestor_request() so that device PM QoS type is passed to it as the third argument and make it support the DEV_PM_QOS_LATENCY_TOLERANCE device PM QoS type (in addition to DEV_PM_QOS_RESUME_LATENCY).
That will allow the drivers of devices without latency tolerance hardware support to use their ancestors having it as proxies for their latency tolerance requirements.
--- Documentation/power/pm_qos_interface.txt | 6 ++++-- drivers/base/power/qos.c | 22 +++++++++++++++++----- drivers/input/touchscreen/st1232.c | 3 ++- include/linux/pm_qos.h | 7 +++++-- 4 files changed, 28 insertions(+), 10 deletions(-)
include/linux/pm_qos.h +5 -2
@@ -148,7 +148,8 @@ int dev_pm_qos_remove_global_notifier(stvoid dev_pm_qos_constraints_init (struct device *dev);void dev_pm_qos_constraints_destroy (struct device *dev);int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value); +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value);#else static inline enum pm_qos_flags_status __dev_pm_qos_flags (struct device *dev,s32 mask)
@@ -191,7 +192,9 @@ static inline void dev_pm_qos_constraint dev->power.power_state = PMSG_INVALID; }static inline int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type, +s32 value) {return 0 ; }#endif
drivers/base/power/qos.c +17 -5
@@ -558,20 +558,32 @@ EXPORT_SYMBOL_GPL(dev_pm_qos_remove_glob* dev_pm_qos_add_ancestor_request - Add PM QoS request for device's ancestor. * @dev: Device whose ancestor to add the request for. * @req: Pointer to the preallocated handle. +* @type: Type of the request. * @value: Constraint latency value. */ int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value) {struct device *ancestor = dev->parent;int ret = -ENODEV ; -while (ancestor && !ancestor->power.ignore_children) - ancestor = ancestor->parent; +switch (type) { +case DEV_PM_QOS_RESUME_LATENCY : +while (ancestor && !ancestor->power.ignore_children) + ancestor = ancestor->parent; +break ; +case DEV_PM_QOS_LATENCY_TOLERANCE : +while (ancestor && !ancestor->power.set_latency_tolerance) + ancestor = ancestor->parent; + +break ; +default : + ancestor =NULL ; + }if (ancestor) - ret =dev_pm_qos_add_request (ancestor, req, -DEV_PM_QOS_RESUME_LATENCY , value); + ret =dev_pm_qos_add_request (ancestor, req, type, value);if (ret <0 ) req->dev =NULL ;
Documentation/power/pm_qos_interface.txt +4 -2
@@ -134,9 +134,11 @@ The meaning of the return values is as f PM_QOS_FLAGS_UNDEFINED: The device's PM QoS structure has not been initialized or the list of requests is empty. -int dev_pm_qos_add_ancestor_request(dev, handle, value) +int dev_pm_qos_add_ancestor_request(dev, handle, type, value) Add a PM QoS request for the first direct ancestor of the given device whose -power.ignore_children flag is unset. +power.ignore_children flag is unset (for DEV_PM_QOS_RESUME_LATENCY requests) +or whose power.set_latency_tolerance callback pointer is not NULL (for +DEV_PM_QOS_LATENCY_TOLERANCE requests). int dev_pm_qos_expose_latency_limit(device, value) Add a request to the device's PM QoS list of resume latency constraints and
drivers/input/touchscreen/st1232.c +2 -1
@@ -134,7 +134,8 @@ static irqreturn_t st1232_ts_irq_handler }else if (!ts->low_latency_req.dev) {/* First contact, request 100 us latency. */ dev_pm_qos_add_ancestor_request (&ts->client->dev, - &ts->low_latency_req,100 ); + &ts->low_latency_req, +DEV_PM_QOS_RESUME_LATENCY ,100 ); }/* SYN_REPORT */
v2 cover PM / QoS: Introduce latency tolerance device PM QoS type
In reply to Rafael J. WysockiQuoted 28 lines · Rafael J. Wysocki
Hi All,
On some platforms hardware may switch to an energy-saving mode on the fly on the basis of certain utilization metrics used by it. That usually is desirable from the energy conservation standpoint, but it generally causes latencies to increase which may adversely affect some operations. For this reason, the platforms in question usually provide some interfaces for software to indicate its latency tolerance and possibly to prevent the energy-saving modes from being selected too aggressively.
The following series of patches introduces a device PM QoS type allowing those interfaces to be used by kernel code and user space. It is designed in analogy with the existing resume latency device PM QoS type, which allows some pieces of the existing device PM QoS code to be re-used and makes the new user space interface fit into the existing framework.
Patch [1/5] modifies the names of symbols, variables, functions and structure fields associated with the existing resume latency device PM QoS type to avoid any confusion with the new one introduced by the subsequent patches.
Patch [2/5] introduces a new field in struct
pm_qos_constraintsfor specifying a special value to be returned as the effective requirement when the given list of PM QoS requirements is empty. That field is necessary for the new latency tolerance device PM QoS type.Patch [3/5] introduces the latency tolerance device PM QoS type along with documentation.
Quoted 6 lines
Patch [4/5] modifies the ACPI LPSS (Low-Power Subsystem) driver to hook up LPSS devices to the new latency tolerance device PM QoS interface.
Patch [5/5] modifies the
dev_pm_qos_add_ancestor_request()routine so that it can be used by drivers of devices without hardware latency tolerance support for specifying their requirements via the ancestors of those devies.
As usual, testing uncovered some issues, so an updated series follows.
Thanks!
Signature · 2 lines
I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
PATCH v2 1/6 PM / QoS: Rename device resume latency QoS items
From: Rafael J. Wysocki <[email protected]>
Rename symbols, variables, functions and structure fields related do the resume latency device PM QoS type so that it is clear where they belong (in particular, to avoid confusion with the latency tolerance device PM QoS type introduced by a subsequent changeset).
Update the PM QoS documentation to better reflect its current state.
--- Documentation/power/pm_qos_interface.txt | 37 +++++++++----------- Documentation/trace/events-power.txt | 2 - drivers/base/power/power.h | 4 +- drivers/base/power/qos.c | 55 +++++++++++++++---------------- drivers/base/power/sysfs.c | 32 +++++++++--------- drivers/mtd/nand/sh_flctl.c | 2 - include/linux/pm_qos.h | 14 +++---- include/trace/events/power.h | 4 +- 8 files changed, 73 insertions(+), 77 deletions(-)
Documentation/trace/events-power.txt +1 -1
@@ -92,5 +92,5 @@ dev_pm_qos_remove_request "devi The first parameter gives the device name which tries to add/update/remove QoS requests. -The second parameter gives the request type (e.g. "DEV_PM_QOS_LATENCY"). +The second parameter gives the request type (e.g. "DEV_PM_QOS_RESUME_LATENCY"). The third parameter is value to be added/updated/removed.
drivers/base/power/qos.c +27 -28
151 lines not on this page · raw message
drivers/mtd/nand/sh_flctl.c +1 -1
@@ -897,7 +897,7 @@ static void flctl_select_chip(struct mtdif (!flctl->qos_request) { ret =dev_pm_qos_add_request (&flctl->pdev->dev, &flctl->pm_qos, -DEV_PM_QOS_LATENCY , +DEV_PM_QOS_RESUME_LATENCY ,100 );if (ret <0 ) dev_err(&flctl->pdev->dev,
include/linux/pm_qos.h +7 -7
@@ -32,7 +32,7 @@ enum pm_qos_flags_status {#define PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0 -#define PM_QOS_DEV_LAT_DEFAULT_VALUE 0 +#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -49,7 +49,7 @@ struct pm_qos_flags_request { };enum dev_pm_qos_req_type { -DEV_PM_QOS_LATENCY =1 , +DEV_PM_QOS_RESUME_LATENCY =1 ,DEV_PM_QOS_FLAGS , };
@@ -87,9 +87,9 @@ struct pm_qos_flags { };struct dev_pm_qos { -struct pm_qos_constraints latency; +struct pm_qos_constraints resume_latency;struct pm_qos_flags flags; -struct dev_pm_qos_request *latency_req; +struct dev_pm_qos_request *resume_latency_req;struct dev_pm_qos_request *flags_req; };
@@ -196,9 +196,9 @@ int dev_pm_qos_expose_flags(struct devicvoid dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); -static inline s32 dev_pm_qos_requested_latency (struct device *dev) +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) { -return dev->power.qos->latency_req->data.pnode.prio; +return dev->power.qos->resume_latency_req->data.pnode.prio; }static inline s32 dev_pm_qos_requested_flags (struct device *dev)
@@ -215,7 +215,7 @@ static inline void dev_pm_qos_hide_flagsstatic inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } -static inline s32 dev_pm_qos_requested_latency (struct device *dev) {return 0 ; } +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }#endif
include/trace/events/power.h +2 -2
@@ -359,8 +359,8 @@ DECLARE_EVENT_CLASS(dev_pm_qos_request,TP_printk ("device=%s type=%s new_value=%d" ,__get_str (name),__print_symbolic (__entry->type, - { DEV_PM_QOS_LATENCY, "DEV_PM_QOS_LATENCY" }, - { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), + { DEV_PM_QOS_RESUME_LATENCY, "DEV_PM_QOS_RESUME_LATENCY" }, + { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), __entry->new_value) );
drivers/base/power/power.h +2 -2
@@ -89,8 +89,8 @@ extern void dpm_sysfs_remove(struct deviextern void rpm_sysfs_remove (struct device *dev);extern int wakeup_sysfs_add (struct device *dev);extern void wakeup_sysfs_remove (struct device *dev); -extern int pm_qos_sysfs_add_latency (struct device *dev); -extern void pm_qos_sysfs_remove_latency (struct device *dev); +extern int pm_qos_sysfs_add_resume_latency (struct device *dev); +extern void pm_qos_sysfs_remove_resume_latency (struct device *dev);extern int pm_qos_sysfs_add_flags (struct device *dev);extern void pm_qos_sysfs_remove_flags (struct device *dev);
drivers/base/power/sysfs.c +17 -15
@@ -218,15 +218,16 @@ static ssize_t autosuspend_delay_ms_stor static DEVICE_ATTR(autosuspend_delay_ms, 0644, autosuspend_delay_ms_show, autosuspend_delay_ms_store); -static ssize_t pm_qos_latency_show (struct device *dev, -struct device_attribute *attr,char *buf) +static ssize_t pm_qos_resume_latency_show (struct device *dev, +struct device_attribute *attr, +char *buf) { -return sprintf (buf,"%d\n" ,dev_pm_qos_requested_latency (dev)); +return sprintf (buf,"%d\n" ,dev_pm_qos_requested_resume_latency (dev)); } -static ssize_t pm_qos_latency_store (struct device *dev, -struct device_attribute *attr, -const char *buf,size_t n) +static ssize_t pm_qos_resume_latency_store (struct device *dev, +struct device_attribute *attr, +const char *buf,size_t n) {s32 value;int ret;
@@ -237,12 +238,13 @@ static ssize_t pm_qos_latency_store(struif (value <0 )return -EINVAL ; - ret =dev_pm_qos_update_request (dev->power.qos->latency_req, value); + ret =dev_pm_qos_update_request (dev->power.qos->resume_latency_req, + value);return ret <0 ? ret : n; } static DEVICE_ATTR(pm_qos_resume_latency_us, 0644, - pm_qos_latency_show, pm_qos_latency_store); + pm_qos_resume_latency_show, pm_qos_resume_latency_store); static ssize_t pm_qos_no_power_off_show(struct device *dev, struct device_attribute *attr,
@@ -618,15 +620,15 @@ static struct attribute_group pm_runtime .attrs = runtime_attrs, }; -static struct attribute *pm_qos_latency_attrs[] = { +static struct attribute *pm_qos_resume_latency_attrs[] = { #ifdef CONFIG_PM_RUNTIME &dev_attr_pm_qos_resume_latency_us.attr, #endif /* CONFIG_PM_RUNTIME */NULL , }; -static struct attribute_group pm_qos_latency_attr_group = { +static struct attribute_group pm_qos_resume_latency_attr_group = { .name = power_group_name, - .attrs = pm_qos_latency_attrs, + .attrs = pm_qos_resume_latency_attrs, };static struct attribute *pm_qos_flags_attrs[] = {
@@ -681,14 +683,14 @@ void wakeup_sysfs_remove(struct device *sysfs_unmerge_group (&dev->kobj, &pm_wakeup_attr_group); } -int pm_qos_sysfs_add_latency (struct device *dev) +int pm_qos_sysfs_add_resume_latency (struct device *dev) { -return sysfs_merge_group (&dev->kobj, &pm_qos_latency_attr_group); +return sysfs_merge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); } -void pm_qos_sysfs_remove_latency (struct device *dev) +void pm_qos_sysfs_remove_resume_latency (struct device *dev) { -sysfs_unmerge_group (&dev->kobj, &pm_qos_latency_attr_group); +sysfs_unmerge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); }int pm_qos_sysfs_add_flags (struct device *dev)
Documentation/power/pm_qos_interface.txt +16 -21
@@ -89,16 +89,16 @@ node. 2. PM QoS per-device latency and flags framework For each device, there are two lists of PM QoS requests. One is maintained -along with the aggregated target of latency value and the other is for PM QoS -flags. Values are updated in response to changes of the request list. +along with the aggregated target of resume latency value and the other is for +PM QoS flags. Values are updated in response to changes of the request list. -Target latency value is simply the minimum of the request values held in the -parameter list elements. The PM QoS flags aggregate value is a gather (bitwise -OR) of all list elements' values. Two device PM QoS flags are defined currently: -PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. +Target resume latency value is simply the minimum of the request values held in +the parameter list elements. The PM QoS flags aggregate value is a gather +(bitwise OR) of all list elements' values. Two device PM QoS flags are defined +currently: PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. -Note: the aggregated target value is implemented as an atomic variable so that -reading the aggregated value does not require any locking mechanism. +Note: the aggregated target value is implemented in such a way that reading the +aggregated value does not require any locking mechanism. From kernel mode the use of this interface is the following:
@@ -137,14 +137,14 @@ Add a PM QoS request for the first direc power.ignore_children flag is unset. int dev_pm_qos_expose_latency_limit(device, value) -Add a request to the device's PM QoS list of latency constraints and create -a sysfs attribute pm_qos_resume_latency_us under the device's power directory -allowing user space to manipulate that request. +Add a request to the device's PM QoS list of resume latency constraints and +create a sysfs attribute pm_qos_resume_latency_us under the device's power +directory allowing user space to manipulate that request. void dev_pm_qos_hide_latency_limit(device) Drop the request added by dev_pm_qos_expose_latency_limit() from the device's -PM QoS list of latency constraints and remove sysfs attribute pm_qos_resume_latency_us -from the device's power directory. +PM QoS list of resume latency constraints and remove sysfs attribute +pm_qos_resume_latency_us from the device's power directory. int dev_pm_qos_expose_flags(device, value) Add a request to the device's PM QoS list of flags and create sysfs attributes
@@ -163,7 +163,7 @@ a per-device notification tree and a glo int dev_pm_qos_add_notifier(device, notifier): Adds a notification callback function for the device. The callback is called when the aggregated value of the device constraints list -is changed. +is changed (for resume latency device PM QoS only). int dev_pm_qos_remove_notifier(device, notifier): Removes the notification callback function for the device.
@@ -171,14 +171,9 @@ Removes the notification callback functi int dev_pm_qos_add_global_notifier(notifier): Adds a notification callback function in the global notification tree of the framework. -The callback is called when the aggregated value for any device is changed. +The callback is called when the aggregated value for any device is changed +(for resume latency device PM QoS only). int dev_pm_qos_remove_global_notifier(notifier): Removes the notification callback function from the global notification tree of the framework. - - -From user mode: -No API for user space access to the per-device latency constraints is provided -yet - still under discussion. -
PATCH v2 2/6 PM / QoS: Add no_constraints_value field to struct pm_qos_constraints
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new field, no_constraints_value, to struct pm_qos_constraints representing a list of PM QoS constraint requests to be returned by pm_qos_get_value() when that list of requests is empty.
That field will be equal to default_value for all of the existing global PM QoS classes and for the resume latency device PM QoS type, but it will be different from default_value for the new latency tolerance device PM QoS type introduced by the next changeset.
--- drivers/base/power/qos.c | 1 + include/linux/pm_qos.h | 1 + kernel/power/qos.c | 5 ++++- 3 files changed, 6 insertions(+), 1 deletion(-)
drivers/base/power/qos.c +1 -0
@@ -190,6 +190,7 @@ static int dev_pm_qos_constraints_allocaplist_head_init (&c->list); c->target_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->default_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; + c->no_constraint_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->type =PM_QOS_MIN ; c->notifiers = n;
include/linux/pm_qos.h +1 -0
@@ -77,6 +77,7 @@ struct pm_qos_constraints {struct plist_head list;s32 target_value;/* Do not change to 64 bit */ s32 default_value; +s32 no_constraint_value;enum pm_qos_type type;struct blocking_notifier_head *notifiers; };
kernel/power/qos.c +4 -1
@@ -66,6 +66,7 @@ static struct pm_qos_constraints cpu_dma .list = PLIST_HEAD_INIT(cpu_dma_constraints.list), .target_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .default_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &cpu_dma_lat_notifier, };
@@ -79,6 +80,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_lat_constraints.list), .target_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &network_lat_notifier, };
@@ -93,6 +95,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_tput_constraints.list), .target_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .type = PM_QOS_MAX, .notifiers = &network_throughput_notifier, };
@@ -128,7 +131,7 @@ static const struct file_operations pm_qstatic inline int pm_qos_get_value (struct pm_qos_constraints *c) {if (plist_head_empty (&c->list)) -return c->default_value; +return c->no_constraint_value;switch (c->type) {case PM_QOS_MIN :
PATCH v2 3/6 PM / QoS: Introcuce latency tolerance device PM QoS type
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new latency tolerance device PM QoS type to be use for specifying active state (RPM_ACTIVE) memory access (DMA) latency tolerance requirements for devices. It may be used to prevent hardware from choosing overly aggressive energy-saving operation modes (causing too much latency to appear) for the whole platform.
This feature reqiures hardware support, so it only will be available for devices having a new .set_latency_tolerance() callback in struct dev_pm_info populated, in which case the routine pointed to by it should implement whatever is necessary to transfer the effective requirement value to the hardware.
Whenever the effective latency tolerance changes for the device, its .set_latency_tolerance() callback will be executed and the effective value will be passed to it. If that value is negative, which means that the list of latency tolerance requirements for the device is empty, the callback is expected to switch the underlying hardware latency tolerance control mechanism to an autonomous mode if available. If that value is 0, in turn, and the hardware supports a special "no requirement" setting, the callback is expected to use it. That allows software to prevent the hardware from automatically updating the device's latency tolerance in response to its power state changes (e.g. during transitions from D3cold to D0), which generally may be done in the autonomous latency tolerance control mode.
If .set_latency_tolerance() is present for the device, a new pm_qos_latency_tolerance_us attribute will be present in the devivce's power directory in sysfs. Then, user space can use that attribute to specify its latency tolerance requirement for the device, if any. Writing 0 to it means "no requirement, but do not let the hardware control latency tolerance" and using a negative value allows the hardware to be switched to the autonomous mode if there are no other requirements from the kernel side in the device's list.
This changeset includes a fix from Mika Westerberg.
--- Documentation/ABI/testing/sysfs-devices-power | 25 ++++ Documentation/power/pm_qos_interface.txt | 61 +++++++++-- drivers/base/power/qos.c | 144 ++++++++++++++++++++++---- drivers/base/power/sysfs.c | 56 ++++++++-- include/linux/pm.h | 1 include/linux/pm_qos.h | 11 + kernel/power/qos.c | 13 +- 7 files changed, 266 insertions(+), 45 deletions(-)
include/linux/pm_qos.h +11 -0
@@ -33,6 +33,8 @@ enum pm_qos_flags_status {#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT (-1)#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -50,6 +52,7 @@ struct pm_qos_flags_request {enum dev_pm_qos_req_type {DEV_PM_QOS_RESUME_LATENCY =1 , +DEV_PM_QOS_LATENCY_TOLERANCE ,DEV_PM_QOS_FLAGS , };
@@ -89,8 +92,10 @@ struct pm_qos_flags {struct dev_pm_qos {struct pm_qos_constraints resume_latency; +struct pm_qos_constraints latency_tolerance;struct pm_qos_flags flags;struct dev_pm_qos_request *resume_latency_req; +struct dev_pm_qos_request *latency_tolerance_req;struct dev_pm_qos_request *flags_req; };
@@ -196,6 +201,8 @@ void dev_pm_qos_hide_latency_limit(strucint dev_pm_qos_expose_flags (struct device *dev,s32 value);void dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); +s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev); +int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val);static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {
@@ -215,6 +222,10 @@ static inline int dev_pm_qos_expose_flagstatic inline void dev_pm_qos_hide_flags (struct device *dev) {}static inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } +static inline s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev) + {return PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT ; } +static inline int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val) + {return 0 ; }static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }
drivers/base/power/qos.c +122 -22
195 lines not on this page · raw message
drivers/base/power/sysfs.c +49 -7
85 lines not on this page · raw message
include/linux/pm.h +1 -0
@@ -583,6 +583,7 @@ struct dev_pm_info {unsigned long accounting_timestamp;#endif struct pm_subsys_data *subsys_data;/* Owned by the subsystem. */ +void (*set_latency_tolerance)(struct device *,s32 );struct dev_pm_qos *qos; };
kernel/power/qos.c +8 -5
@@ -173,6 +173,7 @@ int pm_qos_update_target(struct pm_qos_c {unsigned long flags;int prev_value, curr_value, new_value; +int ret;spin_lock_irqsave (&pm_qos_lock, flags); prev_value =pm_qos_get_value (c);
@@ -208,13 +209,15 @@ int pm_qos_update_target(struct pm_qos_ctrace_pm_qos_update_target (action, prev_value, curr_value);if (prev_value != curr_value) { -blocking_notifier_call_chain (c->notifiers, - (unsigned long )curr_value, -NULL ); -return 1 ; + ret =1 ; +if (c->notifiers) +blocking_notifier_call_chain (c->notifiers, + (unsigned long )curr_value, +NULL ); }else { -return 0 ; + ret =0 ; } +return ret; }/**
Documentation/power/pm_qos_interface.txt +51 -10
71 lines not on this page · raw message
Documentation/ABI/testing/sysfs-devices-power +24 -1
@@ -187,7 +187,7 @@ Description: Not all drivers support this attribute. If it isn't supported, attempts to read or write it will yield I/O errors. -What: /sys/devices/.../power/pm_qos_latency_us +What: /sys/devices/.../power/pm_qos_resume_latency_us Date: March 2012 Contact: Rafael J. Wysocki <[email protected]> Description:
@@ -205,6 +205,29 @@ Description: This attribute has no effect on system-wide suspend/resume and hibernation. +What: /sys/devices/.../power/pm_qos_latency_tolerance_us +Date: January 2014 +Contact: Rafael J. Wysocki <[email protected]> +Description: + The /sys/devices/.../power/pm_qos_latency_tolerance_us attribute + contains the PM QoS active state latency tolerance limit for the + given device, which is the maximum memory access latency the + device can suffer without any visible adverse effects on user + space functionality. If that value is 0, the latency does not + matter to user space at all. + + Reading "n/a" from this file means that the maximum memory + access latency for the device may be determined automatically + by the hardware as needed. Writing a negative value to it + allows the hardware to be switched to this mode if there are no + other latency tolerance requirements from the kernel side. + + This attribute is only present if the feature controlled by it + is supported by the hardware. + + This attribute has no effect on runtime suspend and resume of + devices and on system-wide suspend/resume and hibernation. + What: /sys/devices/.../power/pm_qos_no_power_off Date: September 2012 Contact: Rafael J. Wysocki <[email protected]>
PATCH v2 3/6 PM / QoS: Introcuce latency tolerance device PM QoS type
From: Rafael J. Wysocki <[email protected]>
Add a new latency tolerance device PM QoS type to be use for specifying active state (RPM_ACTIVE) memory access (DMA) latency tolerance requirements for devices. It may be used to prevent hardware from choosing overly aggressive energy-saving operation modes (causing too much latency to appear) for the whole platform.
This feature reqiures hardware support, so it only will be available for devices having a new .set_latency_tolerance() callback in struct dev_pm_info populated, in which case the routine pointed to by it should implement whatever is necessary to transfer the effective requirement value to the hardware.
Whenever the effective latency tolerance changes for the device, its .set_latency_tolerance() callback will be executed and the effective value will be passed to it. If that value is negative, which means that the list of latency tolerance requirements for the device is empty, the callback is expected to switch the underlying hardware latency tolerance control mechanism to an autonomous mode if available. If that value is PM_QOS_LATENCY_ANY, in turn, and the hardware supports a special "no requirement" setting, the callback is expected to use it. That allows software to prevent the hardware from automatically updating the device's latency tolerance in response to its power state changes (e.g. during transitions from D3cold to D0), which generally may be done in the autonomous latency tolerance control mode.
If .set_latency_tolerance() is present for the device, a new pm_qos_latency_tolerance_us attribute will be present in the devivce's power directory in sysfs. Then, user space can use that attribute to specify its latency tolerance requirement for the device, if any. Writing "any" to it means "no requirement, but do not let the hardware control latency tolerance" and writing "none" to it allows the hardware to be switched to the autonomous mode if there are no other requirements from the kernel side in the device's list.
This changeset includes a fix from Mika Westerberg.
---
In the previous version 0 was a special "no requirement" value, but of course the plist code doesn't treat 0 in any special way, so if one of the requirements in the list was set to 0, it would end up as the effective value and would enforce the "no requirement" setting on everybody (even if someone had a specific requirement with nonzero value).
For this reason, the update introduces a special value PM_QOS_LATENCY_ANY to be used for that, which is just the maximum value of s32, so it should always be at the end of the plist. Still, the sysfs attribute now has two special cases to handle, so I think it's better to represent them both as strings.
The documentation has been updated to follow the code changes.
Thanks,
Rafael
--- Documentation/ABI/testing/sysfs-devices-power | 27 ++++ Documentation/power/pm_qos_interface.txt | 61 +++++++++-- drivers/base/power/qos.c | 144 ++++++++++++++++++++++---- drivers/base/power/sysfs.c | 65 ++++++++++- include/linux/pm.h | 1 include/linux/pm_qos.h | 12 ++ kernel/power/qos.c | 13 +- 7 files changed, 278 insertions(+), 45 deletions(-)
include/linux/pm_qos.h +12 -0
@@ -33,6 +33,9 @@ enum pm_qos_flags_status {#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT (-1) +#define PM_QOS_LATENCY_ANY ((s32)(~(__u32)0 >> 1))#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -50,6 +53,7 @@ struct pm_qos_flags_request {enum dev_pm_qos_req_type {DEV_PM_QOS_RESUME_LATENCY =1 , +DEV_PM_QOS_LATENCY_TOLERANCE ,DEV_PM_QOS_FLAGS , };
@@ -89,8 +93,10 @@ struct pm_qos_flags {struct dev_pm_qos {struct pm_qos_constraints resume_latency; +struct pm_qos_constraints latency_tolerance;struct pm_qos_flags flags;struct dev_pm_qos_request *resume_latency_req; +struct dev_pm_qos_request *latency_tolerance_req;struct dev_pm_qos_request *flags_req; };
@@ -196,6 +202,8 @@ void dev_pm_qos_hide_latency_limit(strucint dev_pm_qos_expose_flags (struct device *dev,s32 value);void dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); +s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev); +int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val);static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {
@@ -215,6 +223,10 @@ static inline int dev_pm_qos_expose_flagstatic inline void dev_pm_qos_hide_flags (struct device *dev) {}static inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } +static inline s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev) + {return PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT ; } +static inline int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val) + {return 0 ; }static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }
drivers/base/power/qos.c +122 -22
195 lines not on this page · raw message
drivers/base/power/sysfs.c +58 -7
94 lines not on this page · raw message
include/linux/pm.h +1 -0
@@ -583,6 +583,7 @@ struct dev_pm_info {unsigned long accounting_timestamp;#endif struct pm_subsys_data *subsys_data;/* Owned by the subsystem. */ +void (*set_latency_tolerance)(struct device *,s32 );struct dev_pm_qos *qos; };
kernel/power/qos.c +8 -5
@@ -173,6 +173,7 @@ int pm_qos_update_target(struct pm_qos_c {unsigned long flags;int prev_value, curr_value, new_value; +int ret;spin_lock_irqsave (&pm_qos_lock, flags); prev_value =pm_qos_get_value (c);
@@ -208,13 +209,15 @@ int pm_qos_update_target(struct pm_qos_ctrace_pm_qos_update_target (action, prev_value, curr_value);if (prev_value != curr_value) { -blocking_notifier_call_chain (c->notifiers, - (unsigned long )curr_value, -NULL ); -return 1 ; + ret =1 ; +if (c->notifiers) +blocking_notifier_call_chain (c->notifiers, + (unsigned long )curr_value, +NULL ); }else { -return 0 ; + ret =0 ; } +return ret; }/**
Documentation/power/pm_qos_interface.txt +51 -10
71 lines not on this page · raw message
Documentation/ABI/testing/sysfs-devices-power +26 -1
@@ -187,7 +187,7 @@ Description: Not all drivers support this attribute. If it isn't supported, attempts to read or write it will yield I/O errors. -What: /sys/devices/.../power/pm_qos_latency_us +What: /sys/devices/.../power/pm_qos_resume_latency_us Date: March 2012 Contact: Rafael J. Wysocki <[email protected]> Description:
@@ -205,6 +205,31 @@ Description: This attribute has no effect on system-wide suspend/resume and hibernation. +What: /sys/devices/.../power/pm_qos_latency_tolerance_us +Date: January 2014 +Contact: Rafael J. Wysocki <[email protected]> +Description: + The /sys/devices/.../power/pm_qos_latency_tolerance_us attribute + contains the PM QoS active state latency tolerance limit for the + given device in microseconds. That is the maximum memory access + latency the device can suffer without any visible adverse + effects on user space functionality. If that value is the + string "any", the latency does not matter to user space at all, + but hardware should not be allowed to set the latency tolerance + for the device automatically. + + Reading "none" from this file means that the maximum memory + access latency for the device may be determined automatically + by the hardware as needed. Writing "none" to it allows the + hardware to be switched to this mode if there are no other + latency tolerance requirements from the kernel side. + + This attribute is only present if the feature controlled by it + is supported by the hardware. + + This attribute has no effect on runtime suspend and resume of + devices and on system-wide suspend/resume and hibernation. + What: /sys/devices/.../power/pm_qos_no_power_off Date: September 2012 Contact: Rafael J. Wysocki <[email protected]>
PATCH v2 4/6 ACPI / scan: Add bind/unbind callbacks to struct acpi_scan_handler
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
In some cases it may be necessary to perform certain setup/cleanup operations on a device object representing a physical device after it has been associated with an ACPI companion by acpi_bind_one() or before disassociating it from that companion by acpi_unbind_one(), respectively. If there is a struct acpi_bus_type object for the given device's bus type, the .setup()/.cleanup() callbacks from there are executed for these purposes. However, an analogous mechanism will be necessary for devices whose bus types don't have corresponding struct acpi_bus_type objects and that have specific ACPI scan handlers.
For those devices, add new .bind() and .unbind() callbacks to struct acpi_scan_handler that will be executed by acpi_platform_notify() right after the given device has been associated with an ACPI comapnion and by acpi_platform_notify_remove() right before calling acpi_unbind_one() for that device, respectively.
To make that work for scan handlers registering new devices in their .attach() callbacks, modify acpi_scan_attach_handler() to set the ACPI device object's handler field before calling .attach() from the scan handler at hand.
This changeset includes a fix from Mika Westerberg.
--- drivers/acpi/glue.c | 12 ++++++++++++ drivers/acpi/scan.c | 9 +++++---- include/acpi/acpi_bus.h | 2 ++ 3 files changed, 19 insertions(+), 4 deletions(-)
drivers/acpi/glue.c +12 -0
@@ -287,6 +287,7 @@ EXPORT_SYMBOL_GPL(acpi_unbind_one);static int acpi_platform_notify (struct device *dev) {struct acpi_bus_type *type =acpi_get_bus_type (dev); +struct acpi_device *adev;int ret; ret =acpi_bind_one (dev,NULL );
@@ -303,9 +304,14 @@ static int acpi_platform_notify(struct dif (ret) goto out; } + adev =ACPI_COMPANION (dev); +if (!adev) + goto out;if (type && type->setup) type->setup (dev); +else if (adev->handler && adev->handler->bind) + adev->handler->bind (dev); out: #if ACPI_GLUE_DEBUG
@@ -324,11 +330,17 @@ static int acpi_platform_notify(struct dstatic int acpi_platform_notify_remove (struct device *dev) { +struct acpi_device *adev =ACPI_COMPANION (dev);struct acpi_bus_type *type; +if (!adev) +return 0 ; + type =acpi_get_bus_type (dev);if (type && type->cleanup) type->cleanup (dev); +else if (adev->handler && adev->handler->unbind) + adev->handler->unbind (dev);acpi_unbind_one (dev);return 0 ;
include/acpi/acpi_bus.h +2 -0
@@ -133,6 +133,8 @@ struct acpi_scan_handler {struct list_head list_node;int (*attach)(struct acpi_device *dev,const struct acpi_device_id *id);void (*detach)(struct acpi_device *dev); +void (*bind)(struct device *phys_dev); +void (*unbind)(struct device *phys_dev);struct acpi_hotplug_profile hotplug; };
drivers/acpi/scan.c +5 -4
@@ -1952,13 +1952,14 @@ static int acpi_scan_attach_handler(stru handler =acpi_scan_match_handler (hwid->id, &devid);if (handler) { + device->handler = handler; ret = handler->attach (device, devid); -if (ret >0 ) { - device->handler = handler; +if (ret >0 )break ; - }else if (ret <0 ) { + + device->handler =NULL ; +if (ret <0 )break ; - } } }return ret;
PATCH v2 5/6 ACPI / LPSS: Support for device latency tolerance PM QoS
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Add .bind()/.unbind() callbacks to lpss_handler to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition and to clear them on device removal, respectively.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
This changeset includes a fix from Mika Westerberg.
--- drivers/acpi/acpi_lpss.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 70 insertions(+), 1 deletion(-)
drivers/acpi/acpi_lpss.c +70 -1
@@ -33,6 +33,12 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -316,6 +322,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -337,7 +354,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -390,6 +407,38 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); +u32 ltr_mode, ltr_val; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; +if (val >LPSS_LTR_MAX_VAL ) + val =LPSS_LTR_MAX_VAL ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US ; +if (val >0 ) + ltr_val |=LPSS_LTR_SNOOP_REQ ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -427,9 +476,29 @@ static struct notifier_block acpi_lpss_n .notifier_call = acpi_lpss_platform_notify, }; +static void acpi_lpss_bind (struct device *dev) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); + +if (!pdata || !pdata->mmio_base) +return ; + +if (pdata->mmio_size >= pdata->dev_desc->prv_offset +LPSS_LTR_SIZE ) + dev->power.set_latency_tolerance = acpi_lpss_set_ltr; +else +dev_err (dev,"MMIO size insufficient to access LTR\n" ); +} + +static void acpi_lpss_unbind (struct device *dev) +{ + dev->power.set_latency_tolerance =NULL ; +} +static struct acpi_scan_handler lpss_handler = { .ids = acpi_lpss_device_ids, .attach = acpi_lpss_create_device, + .bind = acpi_lpss_bind, + .unbind = acpi_lpss_unbind, };void __init acpi_lpss_init (void )
PATCH v2 5/6 ACPI / LPSS: Support for device latency tolerance PM QoS
From: Rafael J. Wysocki <[email protected]>
Subject: ACPI / LPSS: Support for device latency tolerance PM QoS
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Add .bind()/.unbind() callbacks to lpss_handler to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition and to clear them on device removal, respectively.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
This changeset includes a fix from Mika Westerberg.
---
This one needs to be update to handle PM_QOS_LATENCY_ANY to follow the [3/6] update I've just sent.
Thanks,
Rafael
--- drivers/acpi/acpi_lpss.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 72 insertions(+), 1 deletion(-)
drivers/acpi/acpi_lpss.c +72 -1
@@ -18,6 +18,7 @@#include <linux/io.h> #include <linux/platform_device.h> #include <linux/platform_data/clk-lpss.h> +#include <linux/pm_qos.h> #include <linux/pm_runtime.h> #include "internal.h"
@@ -33,6 +34,12 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -316,6 +323,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -337,7 +355,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -390,6 +408,39 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); +u32 ltr_mode, ltr_val; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val ==PM_QOS_LATENCY_ANY ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US ; + val =LPSS_LTR_MAX_VAL ; + }else if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; +if (val >LPSS_LTR_MAX_VAL ) + val =LPSS_LTR_MAX_VAL ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -427,9 +478,29 @@ static struct notifier_block acpi_lpss_n .notifier_call = acpi_lpss_platform_notify, }; +static void acpi_lpss_bind (struct device *dev) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); + +if (!pdata || !pdata->mmio_base) +return ; + +if (pdata->mmio_size >= pdata->dev_desc->prv_offset +LPSS_LTR_SIZE ) + dev->power.set_latency_tolerance = acpi_lpss_set_ltr; +else +dev_err (dev,"MMIO size insufficient to access LTR\n" ); +} + +static void acpi_lpss_unbind (struct device *dev) +{ + dev->power.set_latency_tolerance =NULL ; +} +static struct acpi_scan_handler lpss_handler = { .ids = acpi_lpss_device_ids, .attach = acpi_lpss_create_device, + .bind = acpi_lpss_bind, + .unbind = acpi_lpss_unbind, };void __init acpi_lpss_init (void )
PATCH v2 6/6 PM / QoS: Add type to dev_pm_qos_add_ancestor_request() arguments
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Rework dev_pm_qos_add_ancestor_request() so that device PM QoS type is passed to it as the third argument and make it support the DEV_PM_QOS_LATENCY_TOLERANCE device PM QoS type (in addition to DEV_PM_QOS_RESUME_LATENCY).
That will allow the drivers of devices without latency tolerance hardware support to use their ancestors having it as proxies for their latency tolerance requirements.
--- Documentation/power/pm_qos_interface.txt | 6 ++++-- drivers/base/power/qos.c | 22 +++++++++++++++++----- drivers/input/touchscreen/st1232.c | 3 ++- include/linux/pm_qos.h | 7 +++++-- 4 files changed, 28 insertions(+), 10 deletions(-)
include/linux/pm_qos.h +5 -2
@@ -148,7 +148,8 @@ int dev_pm_qos_remove_global_notifier(stvoid dev_pm_qos_constraints_init (struct device *dev);void dev_pm_qos_constraints_destroy (struct device *dev);int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value); +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value);#else static inline enum pm_qos_flags_status __dev_pm_qos_flags (struct device *dev,s32 mask)
@@ -191,7 +192,9 @@ static inline void dev_pm_qos_constraint dev->power.power_state = PMSG_INVALID; }static inline int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type, +s32 value) {return 0 ; }#endif
drivers/base/power/qos.c +17 -5
@@ -565,20 +565,32 @@ EXPORT_SYMBOL_GPL(dev_pm_qos_remove_glob* dev_pm_qos_add_ancestor_request - Add PM QoS request for device's ancestor. * @dev: Device whose ancestor to add the request for. * @req: Pointer to the preallocated handle. +* @type: Type of the request. * @value: Constraint latency value. */ int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value) {struct device *ancestor = dev->parent;int ret = -ENODEV ; -while (ancestor && !ancestor->power.ignore_children) - ancestor = ancestor->parent; +switch (type) { +case DEV_PM_QOS_RESUME_LATENCY : +while (ancestor && !ancestor->power.ignore_children) + ancestor = ancestor->parent; +break ; +case DEV_PM_QOS_LATENCY_TOLERANCE : +while (ancestor && !ancestor->power.set_latency_tolerance) + ancestor = ancestor->parent; + +break ; +default : + ancestor =NULL ; + }if (ancestor) - ret =dev_pm_qos_add_request (ancestor, req, -DEV_PM_QOS_RESUME_LATENCY , value); + ret =dev_pm_qos_add_request (ancestor, req, type, value);if (ret <0 ) req->dev =NULL ;
Documentation/power/pm_qos_interface.txt +4 -2
@@ -134,9 +134,11 @@ The meaning of the return values is as f PM_QOS_FLAGS_UNDEFINED: The device's PM QoS structure has not been initialized or the list of requests is empty. -int dev_pm_qos_add_ancestor_request(dev, handle, value) +int dev_pm_qos_add_ancestor_request(dev, handle, type, value) Add a PM QoS request for the first direct ancestor of the given device whose -power.ignore_children flag is unset. +power.ignore_children flag is unset (for DEV_PM_QOS_RESUME_LATENCY requests) +or whose power.set_latency_tolerance callback pointer is not NULL (for +DEV_PM_QOS_LATENCY_TOLERANCE requests). int dev_pm_qos_expose_latency_limit(device, value) Add a request to the device's PM QoS list of resume latency constraints and
drivers/input/touchscreen/st1232.c +2 -1
@@ -134,7 +134,8 @@ static irqreturn_t st1232_ts_irq_handler }else if (!ts->low_latency_req.dev) {/* First contact, request 100 us latency. */ dev_pm_qos_add_ancestor_request (&ts->client->dev, - &ts->low_latency_req,100 ); + &ts->low_latency_req, +DEV_PM_QOS_RESUME_LATENCY ,100 ); }/* SYN_REPORT */
Quoted 29 lines · Rafael J. Wysocki
Quoted 28 lines · Rafael J. Wysocki
Hi All,
On some platforms hardware may switch to an energy-saving mode on the fly on the basis of certain utilization metrics used by it. That usually is desirable from the energy conservation standpoint, but it generally causes latencies to increase which may adversely affect some operations. For this reason, the platforms in question usually provide some interfaces for software to indicate its latency tolerance and possibly to prevent the energy-saving modes from being selected too aggressively.
The following series of patches introduces a device PM QoS type allowing those interfaces to be used by kernel code and user space. It is designed in analogy with the existing resume latency device PM QoS type, which allows some pieces of the existing device PM QoS code to be re-used and makes the new user space interface fit into the existing framework.
Patch [1/5] modifies the names of symbols, variables, functions and structure fields associated with the existing resume latency device PM QoS type to avoid any confusion with the new one introduced by the subsequent patches.
Patch [2/5] introduces a new field in struct
pm_qos_constraintsfor specifying a special value to be returned as the effective requirement when the given list of PM QoS requirements is empty. That field is necessary for the new latency tolerance device PM QoS type.Patch [3/5] introduces the latency tolerance device PM QoS type along with documentation.
Quoted 8 lines
Quoted 6 lines
Patch [4/5] modifies the ACPI LPSS (Low-Power Subsystem) driver to hook up LPSS devices to the new latency tolerance device PM QoS interface.
Patch [5/5] modifies the
dev_pm_qos_add_ancestor_request()routine so that it can be used by drivers of devices without hardware latency tolerance support for specifying their requirements via the ancestors of those devies.As usual, testing uncovered some issues, so an updated series follows.
I tested this series on HSWULT and adding/removing QoS LTR requests from both userspace and from a modified i2c-hid driver (calling dev_pm_qos_add_ancestor_request()) and seems to work nicely -- When I read back the HW LTR values they are what is expected.
Quoted 37 lines · Mika Westerberg
Quoted 36 lines · Rafael J. Wysocki
Quoted 34 lines · Rafael J. Wysocki
Hi All,
On some platforms hardware may switch to an energy-saving mode on the fly on the basis of certain utilization metrics used by it. That usually is desirable from the energy conservation standpoint, but it generally causes latencies to increase which may adversely affect some operations. For this reason, the platforms in question usually provide some interfaces for software to indicate its latency tolerance and possibly to prevent the energy-saving modes from being selected too aggressively.
The following series of patches introduces a device PM QoS type allowing those interfaces to be used by kernel code and user space. It is designed in analogy with the existing resume latency device PM QoS type, which allows some pieces of the existing device PM QoS code to be re-used and makes the new user space interface fit into the existing framework.
Patch [1/5] modifies the names of symbols, variables, functions and structure fields associated with the existing resume latency device PM QoS type to avoid any confusion with the new one introduced by the subsequent patches.
Patch [2/5] introduces a new field in struct
pm_qos_constraintsfor specifying a special value to be returned as the effective requirement when the given list of PM QoS requirements is empty. That field is necessary for the new latency tolerance device PM QoS type.Patch [3/5] introduces the latency tolerance device PM QoS type along with documentation.
Patch [4/5] modifies the ACPI LPSS (Low-Power Subsystem) driver to hook up LPSS devices to the new latency tolerance device PM QoS interface.
Patch [5/5] modifies the
dev_pm_qos_add_ancestor_request()routine so that it can be used by drivers of devices without hardware latency tolerance support for specifying their requirements via the ancestors of those devies.
Quoted 8 lines
Quoted 1 line
As usual, testing uncovered some issues, so an updated series follows.
I tested this series on HSWULT and adding/removing QoS LTR requests from both userspace and from a modified i2c-hid driver (calling
dev_pm_qos_add_ancestor_request()) and seems to work nicely -- When I read back the HW LTR values they are what is expected.Tested-by Mika Westerberg <[email protected]>
Great, thanks a lot for doing that!
Rafael
v3 cover PM / QoS: Introduce latency tolerance device PM QoS type
In reply to Rafael J. WysockiQuoted 29 lines · Rafael J. Wysocki
Quoted 28 lines · Rafael J. Wysocki
Hi All,
On some platforms hardware may switch to an energy-saving mode on the fly on the basis of certain utilization metrics used by it. That usually is desirable from the energy conservation standpoint, but it generally causes latencies to increase which may adversely affect some operations. For this reason, the platforms in question usually provide some interfaces for software to indicate its latency tolerance and possibly to prevent the energy-saving modes from being selected too aggressively.
The following series of patches introduces a device PM QoS type allowing those interfaces to be used by kernel code and user space. It is designed in analogy with the existing resume latency device PM QoS type, which allows some pieces of the existing device PM QoS code to be re-used and makes the new user space interface fit into the existing framework.
Patch [1/5] modifies the names of symbols, variables, functions and structure fields associated with the existing resume latency device PM QoS type to avoid any confusion with the new one introduced by the subsequent patches.
Patch [2/5] introduces a new field in struct
pm_qos_constraintsfor specifying a special value to be returned as the effective requirement when the given list of PM QoS requirements is empty. That field is necessary for the new latency tolerance device PM QoS type.Patch [3/5] introduces the latency tolerance device PM QoS type along with documentation.
Quoted 8 lines
Quoted 6 lines
Patch [4/5] modifies the ACPI LPSS (Low-Power Subsystem) driver to hook up LPSS devices to the new latency tolerance device PM QoS interface.
Patch [5/5] modifies the
dev_pm_qos_add_ancestor_request()routine so that it can be used by drivers of devices without hardware latency tolerance support for specifying their requirements via the ancestors of those devies.As usual, testing uncovered some issues, so an updated series follows.
OK, no comments - no objections. Cool! :-)
Time for a non-RFC submission I guess.
The following series is almost the same as the previous one except for patch [5/6] that fixes a bug (LPSS_LTR_SNOOP_REQ has to be set along with LPSS_LTR_SNOOP_LAT_1US) and introduces a cutoff value such that all above it will mean "no requirement".
Thanks!
Signature · 2 lines
I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center.
PATCH v3 1/6 PM / QoS: Rename device resume latency QoS items
From: Rafael J. Wysocki <[email protected]>
Rename symbols, variables, functions and structure fields related do the resume latency device PM QoS type so that it is clear where they belong (in particular, to avoid confusion with the latency tolerance device PM QoS type introduced by a subsequent changeset).
Update the PM QoS documentation to better reflect its current state.
--- Documentation/power/pm_qos_interface.txt | 37 +++++++++----------- Documentation/trace/events-power.txt | 2 - drivers/base/power/power.h | 4 +- drivers/base/power/qos.c | 55 +++++++++++++++---------------- drivers/base/power/sysfs.c | 32 +++++++++--------- drivers/mtd/nand/sh_flctl.c | 2 - include/linux/pm_qos.h | 14 +++---- include/trace/events/power.h | 4 +- 8 files changed, 73 insertions(+), 77 deletions(-)
Documentation/trace/events-power.txt +1 -1
@@ -92,5 +92,5 @@ dev_pm_qos_remove_request "devi The first parameter gives the device name which tries to add/update/remove QoS requests. -The second parameter gives the request type (e.g. "DEV_PM_QOS_LATENCY"). +The second parameter gives the request type (e.g. "DEV_PM_QOS_RESUME_LATENCY"). The third parameter is value to be added/updated/removed.
drivers/base/power/qos.c +27 -28
151 lines not on this page · raw message
drivers/mtd/nand/sh_flctl.c +1 -1
@@ -897,7 +897,7 @@ static void flctl_select_chip(struct mtdif (!flctl->qos_request) { ret =dev_pm_qos_add_request (&flctl->pdev->dev, &flctl->pm_qos, -DEV_PM_QOS_LATENCY , +DEV_PM_QOS_RESUME_LATENCY ,100 );if (ret <0 ) dev_err(&flctl->pdev->dev,
include/linux/pm_qos.h +7 -7
@@ -32,7 +32,7 @@ enum pm_qos_flags_status {#define PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0 -#define PM_QOS_DEV_LAT_DEFAULT_VALUE 0 +#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -49,7 +49,7 @@ struct pm_qos_flags_request { };enum dev_pm_qos_req_type { -DEV_PM_QOS_LATENCY =1 , +DEV_PM_QOS_RESUME_LATENCY =1 ,DEV_PM_QOS_FLAGS , };
@@ -87,9 +87,9 @@ struct pm_qos_flags { };struct dev_pm_qos { -struct pm_qos_constraints latency; +struct pm_qos_constraints resume_latency;struct pm_qos_flags flags; -struct dev_pm_qos_request *latency_req; +struct dev_pm_qos_request *resume_latency_req;struct dev_pm_qos_request *flags_req; };
@@ -196,9 +196,9 @@ int dev_pm_qos_expose_flags(struct devicvoid dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); -static inline s32 dev_pm_qos_requested_latency (struct device *dev) +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) { -return dev->power.qos->latency_req->data.pnode.prio; +return dev->power.qos->resume_latency_req->data.pnode.prio; }static inline s32 dev_pm_qos_requested_flags (struct device *dev)
@@ -215,7 +215,7 @@ static inline void dev_pm_qos_hide_flagsstatic inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } -static inline s32 dev_pm_qos_requested_latency (struct device *dev) {return 0 ; } +static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }#endif
include/trace/events/power.h +2 -2
@@ -412,8 +412,8 @@ DECLARE_EVENT_CLASS(dev_pm_qos_request,TP_printk ("device=%s type=%s new_value=%d" ,__get_str (name),__print_symbolic (__entry->type, - { DEV_PM_QOS_LATENCY, "DEV_PM_QOS_LATENCY" }, - { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), + { DEV_PM_QOS_RESUME_LATENCY, "DEV_PM_QOS_RESUME_LATENCY" }, + { DEV_PM_QOS_FLAGS, "DEV_PM_QOS_FLAGS" }), __entry->new_value) );
drivers/base/power/power.h +2 -2
@@ -89,8 +89,8 @@ extern void dpm_sysfs_remove(struct deviextern void rpm_sysfs_remove (struct device *dev);extern int wakeup_sysfs_add (struct device *dev);extern void wakeup_sysfs_remove (struct device *dev); -extern int pm_qos_sysfs_add_latency (struct device *dev); -extern void pm_qos_sysfs_remove_latency (struct device *dev); +extern int pm_qos_sysfs_add_resume_latency (struct device *dev); +extern void pm_qos_sysfs_remove_resume_latency (struct device *dev);extern int pm_qos_sysfs_add_flags (struct device *dev);extern void pm_qos_sysfs_remove_flags (struct device *dev);
drivers/base/power/sysfs.c +17 -15
@@ -218,15 +218,16 @@ static ssize_t autosuspend_delay_ms_stor static DEVICE_ATTR(autosuspend_delay_ms, 0644, autosuspend_delay_ms_show, autosuspend_delay_ms_store); -static ssize_t pm_qos_latency_show (struct device *dev, -struct device_attribute *attr,char *buf) +static ssize_t pm_qos_resume_latency_show (struct device *dev, +struct device_attribute *attr, +char *buf) { -return sprintf (buf,"%d\n" ,dev_pm_qos_requested_latency (dev)); +return sprintf (buf,"%d\n" ,dev_pm_qos_requested_resume_latency (dev)); } -static ssize_t pm_qos_latency_store (struct device *dev, -struct device_attribute *attr, -const char *buf,size_t n) +static ssize_t pm_qos_resume_latency_store (struct device *dev, +struct device_attribute *attr, +const char *buf,size_t n) {s32 value;int ret;
@@ -237,12 +238,13 @@ static ssize_t pm_qos_latency_store(struif (value <0 )return -EINVAL ; - ret =dev_pm_qos_update_request (dev->power.qos->latency_req, value); + ret =dev_pm_qos_update_request (dev->power.qos->resume_latency_req, + value);return ret <0 ? ret : n; } static DEVICE_ATTR(pm_qos_resume_latency_us, 0644, - pm_qos_latency_show, pm_qos_latency_store); + pm_qos_resume_latency_show, pm_qos_resume_latency_store); static ssize_t pm_qos_no_power_off_show(struct device *dev, struct device_attribute *attr,
@@ -618,15 +620,15 @@ static struct attribute_group pm_runtime .attrs = runtime_attrs, }; -static struct attribute *pm_qos_latency_attrs[] = { +static struct attribute *pm_qos_resume_latency_attrs[] = { #ifdef CONFIG_PM_RUNTIME &dev_attr_pm_qos_resume_latency_us.attr, #endif /* CONFIG_PM_RUNTIME */NULL , }; -static struct attribute_group pm_qos_latency_attr_group = { +static struct attribute_group pm_qos_resume_latency_attr_group = { .name = power_group_name, - .attrs = pm_qos_latency_attrs, + .attrs = pm_qos_resume_latency_attrs, };static struct attribute *pm_qos_flags_attrs[] = {
@@ -681,14 +683,14 @@ void wakeup_sysfs_remove(struct device *sysfs_unmerge_group (&dev->kobj, &pm_wakeup_attr_group); } -int pm_qos_sysfs_add_latency (struct device *dev) +int pm_qos_sysfs_add_resume_latency (struct device *dev) { -return sysfs_merge_group (&dev->kobj, &pm_qos_latency_attr_group); +return sysfs_merge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); } -void pm_qos_sysfs_remove_latency (struct device *dev) +void pm_qos_sysfs_remove_resume_latency (struct device *dev) { -sysfs_unmerge_group (&dev->kobj, &pm_qos_latency_attr_group); +sysfs_unmerge_group (&dev->kobj, &pm_qos_resume_latency_attr_group); }int pm_qos_sysfs_add_flags (struct device *dev)
Documentation/power/pm_qos_interface.txt +16 -21
@@ -89,16 +89,16 @@ node. 2. PM QoS per-device latency and flags framework For each device, there are two lists of PM QoS requests. One is maintained -along with the aggregated target of latency value and the other is for PM QoS -flags. Values are updated in response to changes of the request list. +along with the aggregated target of resume latency value and the other is for +PM QoS flags. Values are updated in response to changes of the request list. -Target latency value is simply the minimum of the request values held in the -parameter list elements. The PM QoS flags aggregate value is a gather (bitwise -OR) of all list elements' values. Two device PM QoS flags are defined currently: -PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. +Target resume latency value is simply the minimum of the request values held in +the parameter list elements. The PM QoS flags aggregate value is a gather +(bitwise OR) of all list elements' values. Two device PM QoS flags are defined +currently: PM_QOS_FLAG_NO_POWER_OFF and PM_QOS_FLAG_REMOTE_WAKEUP. -Note: the aggregated target value is implemented as an atomic variable so that -reading the aggregated value does not require any locking mechanism. +Note: the aggregated target value is implemented in such a way that reading the +aggregated value does not require any locking mechanism. From kernel mode the use of this interface is the following:
@@ -137,14 +137,14 @@ Add a PM QoS request for the first direc power.ignore_children flag is unset. int dev_pm_qos_expose_latency_limit(device, value) -Add a request to the device's PM QoS list of latency constraints and create -a sysfs attribute pm_qos_resume_latency_us under the device's power directory -allowing user space to manipulate that request. +Add a request to the device's PM QoS list of resume latency constraints and +create a sysfs attribute pm_qos_resume_latency_us under the device's power +directory allowing user space to manipulate that request. void dev_pm_qos_hide_latency_limit(device) Drop the request added by dev_pm_qos_expose_latency_limit() from the device's -PM QoS list of latency constraints and remove sysfs attribute pm_qos_resume_latency_us -from the device's power directory. +PM QoS list of resume latency constraints and remove sysfs attribute +pm_qos_resume_latency_us from the device's power directory. int dev_pm_qos_expose_flags(device, value) Add a request to the device's PM QoS list of flags and create sysfs attributes
@@ -163,7 +163,7 @@ a per-device notification tree and a glo int dev_pm_qos_add_notifier(device, notifier): Adds a notification callback function for the device. The callback is called when the aggregated value of the device constraints list -is changed. +is changed (for resume latency device PM QoS only). int dev_pm_qos_remove_notifier(device, notifier): Removes the notification callback function for the device.
@@ -171,14 +171,9 @@ Removes the notification callback functi int dev_pm_qos_add_global_notifier(notifier): Adds a notification callback function in the global notification tree of the framework. -The callback is called when the aggregated value for any device is changed. +The callback is called when the aggregated value for any device is changed +(for resume latency device PM QoS only). int dev_pm_qos_remove_global_notifier(notifier): Removes the notification callback function from the global notification tree of the framework. - - -From user mode: -No API for user space access to the per-device latency constraints is provided -yet - still under discussion. -
PATCH v3 2/6 PM / QoS: Add no_constraints_value field to struct pm_qos_constraints
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new field, no_constraints_value, to struct pm_qos_constraints representing a list of PM QoS constraint requests to be returned by pm_qos_get_value() when that list of requests is empty.
That field will be equal to default_value for all of the existing global PM QoS classes and for the resume latency device PM QoS type, but it will be different from default_value for the new latency tolerance device PM QoS type introduced by the next changeset.
--- drivers/base/power/qos.c | 1 + include/linux/pm_qos.h | 1 + kernel/power/qos.c | 5 ++++- 3 files changed, 6 insertions(+), 1 deletion(-)
drivers/base/power/qos.c +1 -0
@@ -190,6 +190,7 @@ static int dev_pm_qos_constraints_allocaplist_head_init (&c->list); c->target_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->default_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; + c->no_constraint_value =PM_QOS_RESUME_LATENCY_DEFAULT_VALUE ; c->type =PM_QOS_MIN ; c->notifiers = n;
include/linux/pm_qos.h +1 -0
@@ -77,6 +77,7 @@ struct pm_qos_constraints {struct plist_head list;s32 target_value;/* Do not change to 64 bit */ s32 default_value; +s32 no_constraint_value;enum pm_qos_type type;struct blocking_notifier_head *notifiers; };
kernel/power/qos.c +4 -1
@@ -66,6 +66,7 @@ static struct pm_qos_constraints cpu_dma .list = PLIST_HEAD_INIT(cpu_dma_constraints.list), .target_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .default_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_CPU_DMA_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &cpu_dma_lat_notifier, };
@@ -79,6 +80,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_lat_constraints.list), .target_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_LAT_DEFAULT_VALUE, .type = PM_QOS_MIN, .notifiers = &network_lat_notifier, };
@@ -93,6 +95,7 @@ static struct pm_qos_constraints network .list = PLIST_HEAD_INIT(network_tput_constraints.list), .target_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .default_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, + .no_constraint_value = PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE, .type = PM_QOS_MAX, .notifiers = &network_throughput_notifier, };
@@ -128,7 +131,7 @@ static const struct file_operations pm_qstatic inline int pm_qos_get_value (struct pm_qos_constraints *c) {if (plist_head_empty (&c->list)) -return c->default_value; +return c->no_constraint_value;switch (c->type) {case PM_QOS_MIN :
PATCH v3 3/6 PM / QoS: Introcuce latency tolerance device PM QoS type
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new latency tolerance device PM QoS type to be use for specifying active state (RPM_ACTIVE) memory access (DMA) latency tolerance requirements for devices. It may be used to prevent hardware from choosing overly aggressive energy-saving operation modes (causing too much latency to appear) for the whole platform.
This feature reqiures hardware support, so it only will be available for devices having a new .set_latency_tolerance() callback in struct dev_pm_info populated, in which case the routine pointed to by it should implement whatever is necessary to transfer the effective requirement value to the hardware.
Whenever the effective latency tolerance changes for the device, its .set_latency_tolerance() callback will be executed and the effective value will be passed to it. If that value is negative, which means that the list of latency tolerance requirements for the device is empty, the callback is expected to switch the underlying hardware latency tolerance control mechanism to an autonomous mode if available. If that value is PM_QOS_LATENCY_ANY, in turn, and the hardware supports a special "no requirement" setting, the callback is expected to use it. That allows software to prevent the hardware from automatically updating the device's latency tolerance in response to its power state changes (e.g. during transitions from D3cold to D0), which generally may be done in the autonomous latency tolerance control mode.
If .set_latency_tolerance() is present for the device, a new pm_qos_latency_tolerance_us attribute will be present in the devivce's power directory in sysfs. Then, user space can use that attribute to specify its latency tolerance requirement for the device, if any. Writing "any" to it means "no requirement, but do not let the hardware control latency tolerance" and writing "auto" to it allows the hardware to be switched to the autonomous mode if there are no other requirements from the kernel side in the device's list.
This changeset includes a fix from Mika Westerberg.
--- Documentation/ABI/testing/sysfs-devices-power | 27 ++++ Documentation/power/pm_qos_interface.txt | 61 +++++++++-- drivers/base/power/qos.c | 144 ++++++++++++++++++++++---- drivers/base/power/sysfs.c | 65 ++++++++++- include/linux/pm.h | 1 include/linux/pm_qos.h | 12 ++ kernel/power/qos.c | 13 +- 7 files changed, 278 insertions(+), 45 deletions(-)
include/linux/pm_qos.h +12 -0
@@ -33,6 +33,9 @@ enum pm_qos_flags_status {#define PM_QOS_NETWORK_LAT_DEFAULT_VALUE (2000 * USEC_PER_SEC)#define PM_QOS_NETWORK_THROUGHPUT_DEFAULT_VALUE 0#define PM_QOS_RESUME_LATENCY_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_DEFAULT_VALUE 0 +#define PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT (-1) +#define PM_QOS_LATENCY_ANY ((s32)(~(__u32)0 >> 1))#define PM_QOS_FLAG_NO_POWER_OFF (1 << 0)#define PM_QOS_FLAG_REMOTE_WAKEUP (1 << 1)
@@ -50,6 +53,7 @@ struct pm_qos_flags_request {enum dev_pm_qos_req_type {DEV_PM_QOS_RESUME_LATENCY =1 , +DEV_PM_QOS_LATENCY_TOLERANCE ,DEV_PM_QOS_FLAGS , };
@@ -89,8 +93,10 @@ struct pm_qos_flags {struct dev_pm_qos {struct pm_qos_constraints resume_latency; +struct pm_qos_constraints latency_tolerance;struct pm_qos_flags flags;struct dev_pm_qos_request *resume_latency_req; +struct dev_pm_qos_request *latency_tolerance_req;struct dev_pm_qos_request *flags_req; };
@@ -196,6 +202,8 @@ void dev_pm_qos_hide_latency_limit(strucint dev_pm_qos_expose_flags (struct device *dev,s32 value);void dev_pm_qos_hide_flags (struct device *dev);int dev_pm_qos_update_flags (struct device *dev,s32 mask,bool set); +s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev); +int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val);static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {
@@ -215,6 +223,10 @@ static inline int dev_pm_qos_expose_flagstatic inline void dev_pm_qos_hide_flags (struct device *dev) {}static inline int dev_pm_qos_update_flags (struct device *dev,s32 m,bool set) {return 0 ; } +static inline s32 dev_pm_qos_get_user_latency_tolerance (struct device *dev) + {return PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT ; } +static inline int dev_pm_qos_update_user_latency_tolerance (struct device *dev,s32 val) + {return 0 ; }static inline s32 dev_pm_qos_requested_resume_latency (struct device *dev) {return 0 ; }static inline s32 dev_pm_qos_requested_flags (struct device *dev) {return 0 ; }
drivers/base/power/qos.c +122 -22
195 lines not on this page · raw message
drivers/base/power/sysfs.c +58 -7
94 lines not on this page · raw message
include/linux/pm.h +1 -0
@@ -582,6 +582,7 @@ struct dev_pm_info {unsigned long accounting_timestamp;#endif struct pm_subsys_data *subsys_data;/* Owned by the subsystem. */ +void (*set_latency_tolerance)(struct device *,s32 );struct dev_pm_qos *qos; };
kernel/power/qos.c +8 -5
@@ -173,6 +173,7 @@ int pm_qos_update_target(struct pm_qos_c {unsigned long flags;int prev_value, curr_value, new_value; +int ret;spin_lock_irqsave (&pm_qos_lock, flags); prev_value =pm_qos_get_value (c);
@@ -208,13 +209,15 @@ int pm_qos_update_target(struct pm_qos_ctrace_pm_qos_update_target (action, prev_value, curr_value);if (prev_value != curr_value) { -blocking_notifier_call_chain (c->notifiers, - (unsigned long )curr_value, -NULL ); -return 1 ; + ret =1 ; +if (c->notifiers) +blocking_notifier_call_chain (c->notifiers, + (unsigned long )curr_value, +NULL ); }else { -return 0 ; + ret =0 ; } +return ret; }/**
Documentation/power/pm_qos_interface.txt +51 -10
71 lines not on this page · raw message
Documentation/ABI/testing/sysfs-devices-power +26 -1
@@ -187,7 +187,7 @@ Description: Not all drivers support this attribute. If it isn't supported, attempts to read or write it will yield I/O errors. -What: /sys/devices/.../power/pm_qos_latency_us +What: /sys/devices/.../power/pm_qos_resume_latency_us Date: March 2012 Contact: Rafael J. Wysocki <[email protected]> Description:
@@ -205,6 +205,31 @@ Description: This attribute has no effect on system-wide suspend/resume and hibernation. +What: /sys/devices/.../power/pm_qos_latency_tolerance_us +Date: January 2014 +Contact: Rafael J. Wysocki <[email protected]> +Description: + The /sys/devices/.../power/pm_qos_latency_tolerance_us attribute + contains the PM QoS active state latency tolerance limit for the + given device in microseconds. That is the maximum memory access + latency the device can suffer without any visible adverse + effects on user space functionality. If that value is the + string "any", the latency does not matter to user space at all, + but hardware should not be allowed to set the latency tolerance + for the device automatically. + + Reading "auto" from this file means that the maximum memory + access latency for the device may be determined automatically + by the hardware as needed. Writing "auto" to it allows the + hardware to be switched to this mode if there are no other + latency tolerance requirements from the kernel side. + + This attribute is only present if the feature controlled by it + is supported by the hardware. + + This attribute has no effect on runtime suspend and resume of + devices and on system-wide suspend/resume and hibernation. + What: /sys/devices/.../power/pm_qos_no_power_off Date: September 2012 Contact: Rafael J. Wysocki <[email protected]>
PATCH v3 4/6 ACPI / scan: Add bind/unbind callbacks to struct acpi_scan_handler
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
In some cases it may be necessary to perform certain setup/cleanup operations on a device object representing a physical device after it has been associated with an ACPI companion by acpi_bind_one() or before disassociating it from that companion by acpi_unbind_one(), respectively. If there is a struct acpi_bus_type object for the given device's bus type, the .setup()/.cleanup() callbacks from there are executed for these purposes. However, an analogous mechanism will be necessary for devices whose bus types don't have corresponding struct acpi_bus_type objects and that have specific ACPI scan handlers.
For those devices, add new .bind() and .unbind() callbacks to struct acpi_scan_handler that will be executed by acpi_platform_notify() right after the given device has been associated with an ACPI comapnion and by acpi_platform_notify_remove() right before calling acpi_unbind_one() for that device, respectively.
To make that work for scan handlers registering new devices in their .attach() callbacks, modify acpi_scan_attach_handler() to set the ACPI device object's handler field before calling .attach() from the scan handler at hand.
This changeset includes a fix from Mika Westerberg.
--- drivers/acpi/glue.c | 12 ++++++++++++ drivers/acpi/scan.c | 9 +++++---- include/acpi/acpi_bus.h | 2 ++ 3 files changed, 19 insertions(+), 4 deletions(-)
drivers/acpi/glue.c +12 -0
@@ -287,6 +287,7 @@ EXPORT_SYMBOL_GPL(acpi_unbind_one);static int acpi_platform_notify (struct device *dev) {struct acpi_bus_type *type =acpi_get_bus_type (dev); +struct acpi_device *adev;int ret; ret =acpi_bind_one (dev,NULL );
@@ -303,9 +304,14 @@ static int acpi_platform_notify(struct dif (ret) goto out; } + adev =ACPI_COMPANION (dev); +if (!adev) + goto out;if (type && type->setup) type->setup (dev); +else if (adev->handler && adev->handler->bind) + adev->handler->bind (dev); out: #if ACPI_GLUE_DEBUG
@@ -324,11 +330,17 @@ static int acpi_platform_notify(struct dstatic int acpi_platform_notify_remove (struct device *dev) { +struct acpi_device *adev =ACPI_COMPANION (dev);struct acpi_bus_type *type; +if (!adev) +return 0 ; + type =acpi_get_bus_type (dev);if (type && type->cleanup) type->cleanup (dev); +else if (adev->handler && adev->handler->unbind) + adev->handler->unbind (dev);acpi_unbind_one (dev);return 0 ;
include/acpi/acpi_bus.h +2 -0
@@ -133,6 +133,8 @@ struct acpi_scan_handler {struct list_head list_node;int (*attach)(struct acpi_device *dev,const struct acpi_device_id *id);void (*detach)(struct acpi_device *dev); +void (*bind)(struct device *phys_dev); +void (*unbind)(struct device *phys_dev);struct acpi_hotplug_profile hotplug; };
drivers/acpi/scan.c +5 -4
@@ -2020,13 +2020,14 @@ static int acpi_scan_attach_handler(stru handler =acpi_scan_match_handler (hwid->id, &devid);if (handler) { + device->handler = handler; ret = handler->attach (device, devid); -if (ret >0 ) { - device->handler = handler; +if (ret >0 )break ; - }else if (ret <0 ) { + + device->handler =NULL ; +if (ret <0 )break ; - } } }return ret;
PATCH v3 5/6 ACPI / LPSS: Support for device latency tolerance PM QoS
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Add .bind()/.unbind() callbacks to lpss_handler to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition and to clear them on device removal, respectively.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
This changeset includes a fix from Mika Westerberg.
--- drivers/acpi/acpi_lpss.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 70 insertions(+), 1 deletion(-)
drivers/acpi/acpi_lpss.c +70 -1
@@ -33,6 +33,13 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_SNOOP_LAT_CUTOFF 3000 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -315,6 +322,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -336,7 +354,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -389,6 +407,37 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); +u32 ltr_mode, ltr_val; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val >=LPSS_LTR_SNOOP_LAT_CUTOFF ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US ; + val =LPSS_LTR_MAX_VAL ; + }else if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US |LPSS_LTR_SNOOP_REQ ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -426,9 +475,29 @@ static struct notifier_block acpi_lpss_n .notifier_call = acpi_lpss_platform_notify, }; +static void acpi_lpss_bind (struct device *dev) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); + +if (!pdata || !pdata->mmio_base) +return ; + +if (pdata->mmio_size >= pdata->dev_desc->prv_offset +LPSS_LTR_SIZE ) + dev->power.set_latency_tolerance = acpi_lpss_set_ltr; +else +dev_err (dev,"MMIO size insufficient to access LTR\n" ); +} + +static void acpi_lpss_unbind (struct device *dev) +{ + dev->power.set_latency_tolerance =NULL ; +} +static struct acpi_scan_handler lpss_handler = { .ids = acpi_lpss_device_ids, .attach = acpi_lpss_create_device, + .bind = acpi_lpss_bind, + .unbind = acpi_lpss_unbind, };void __init acpi_lpss_init (void )
PATCH v3 5/6 ACPI / LPSS: Support for device latency tolerance PM QoS
From: Rafael J. Wysocki <[email protected]>
Subject: ACPI / LPSS: Support for device latency tolerance PM QoS
Add a new routine, acpi_lpss_set_ltr(), for setting latency tolerance values for LPSS devices having LTR (Latency Tolerance Reporting) registers. Add .bind()/.unbind() callbacks to lpss_handler to set the LPSS devices' power.set_latency_tolerance callback pointers to acpi_lpss_set_ltr() during device addition and to clear them on device removal, respectively.
That will cause the device latency tolerance PM QoS to work for the devices in question as documented.
This changeset includes a fix from Mika Westerberg.
---
acpi_lpss_bind() has to check pdata->dev_desc->ltr_required or it will enable LTR on systems where it may not work.
--- drivers/acpi/acpi_lpss.c | 71 ++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 70 insertions(+), 1 deletion(-)
drivers/acpi/acpi_lpss.c +70 -1
@@ -33,6 +33,13 @@ ACPI_MODULE_NAME("acpi_lpss");#define LPSS_GENERAL_UART_RTS_OVRD BIT(3)#define LPSS_SW_LTR 0x10#define LPSS_AUTO_LTR 0x14 +#define LPSS_LTR_SNOOP_REQ BIT(15) +#define LPSS_LTR_SNOOP_MASK 0x0000FFFF +#define LPSS_LTR_SNOOP_LAT_1US 0x800 +#define LPSS_LTR_SNOOP_LAT_32US 0xC00 +#define LPSS_LTR_SNOOP_LAT_SHIFT 5 +#define LPSS_LTR_SNOOP_LAT_CUTOFF 3000 +#define LPSS_LTR_MAX_VAL 0x3FF#define LPSS_TX_INT 0x20#define LPSS_TX_INT_MASK BIT(1)
@@ -315,6 +322,17 @@ static int acpi_lpss_create_device(strucreturn ret; } +static u32 __lpss_reg_read (struct lpss_private_data *pdata,unsigned int reg) +{ +return readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} + +static void __lpss_reg_write (u32 val,struct lpss_private_data *pdata, +unsigned int reg) +{ +writel (val, pdata->mmio_base + pdata->dev_desc->prv_offset + reg); +} +static int lpss_reg_read (struct device *dev,unsigned int reg,u32 *val) {struct acpi_device *adev;
@@ -336,7 +354,7 @@ static int lpss_reg_read(struct device * ret = -ENODEV ; goto out; } - *val =readl (pdata->mmio_base + pdata->dev_desc->prv_offset + reg); + *val =__lpss_reg_read (pdata, reg); out:spin_unlock_irqrestore (&dev->power.lock, flags);
@@ -389,6 +407,37 @@ static struct attribute_group lpss_attr_ .name = "lpss_ltr", }; +static void acpi_lpss_set_ltr (struct device *dev,s32 val) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); +u32 ltr_mode, ltr_val; + + ltr_mode =__lpss_reg_read (pdata,LPSS_GENERAL ); +if (val <0 ) { +if (ltr_mode &LPSS_GENERAL_LTR_MODE_SW ) { + ltr_mode &= ~LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +return ; + } + ltr_val =__lpss_reg_read (pdata,LPSS_SW_LTR ) & ~LPSS_LTR_SNOOP_MASK ; +if (val >=LPSS_LTR_SNOOP_LAT_CUTOFF ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US ; + val =LPSS_LTR_MAX_VAL ; + }else if (val >LPSS_LTR_MAX_VAL ) { + ltr_val |=LPSS_LTR_SNOOP_LAT_32US |LPSS_LTR_SNOOP_REQ ; + val >>=LPSS_LTR_SNOOP_LAT_SHIFT ; + }else { + ltr_val |=LPSS_LTR_SNOOP_LAT_1US |LPSS_LTR_SNOOP_REQ ; + } + ltr_val |= val; +__lpss_reg_write (ltr_val, pdata,LPSS_SW_LTR ); +if (!(ltr_mode &LPSS_GENERAL_LTR_MODE_SW )) { + ltr_mode |=LPSS_GENERAL_LTR_MODE_SW ; +__lpss_reg_write (ltr_mode, pdata,LPSS_GENERAL ); + } +} +static int acpi_lpss_platform_notify (struct notifier_block *nb,unsigned long action,void *data) {
@@ -426,9 +475,29 @@ static struct notifier_block acpi_lpss_n .notifier_call = acpi_lpss_platform_notify, }; +static void acpi_lpss_bind (struct device *dev) +{ +struct lpss_private_data *pdata =acpi_driver_data (ACPI_COMPANION (dev)); + +if (!pdata || !pdata->mmio_base || !pdata->dev_desc->ltr_required) +return ; + +if (pdata->mmio_size >= pdata->dev_desc->prv_offset +LPSS_LTR_SIZE ) + dev->power.set_latency_tolerance = acpi_lpss_set_ltr; +else +dev_err (dev,"MMIO size insufficient to access LTR\n" ); +} + +static void acpi_lpss_unbind (struct device *dev) +{ + dev->power.set_latency_tolerance =NULL ; +} +static struct acpi_scan_handler lpss_handler = { .ids = acpi_lpss_device_ids, .attach = acpi_lpss_create_device, + .bind = acpi_lpss_bind, + .unbind = acpi_lpss_unbind, };void __init acpi_lpss_init (void )
PATCH v3 6/6 PM / QoS: Add type to dev_pm_qos_add_ancestor_request() arguments
In reply to Rafael J. WysockiFrom: Rafael J. Wysocki <[email protected]>
Rework dev_pm_qos_add_ancestor_request() so that device PM QoS type is passed to it as the third argument and make it support the DEV_PM_QOS_LATENCY_TOLERANCE device PM QoS type (in addition to DEV_PM_QOS_RESUME_LATENCY).
That will allow the drivers of devices without latency tolerance hardware support to use their ancestors having it as proxies for their latency tolerance requirements.
--- Documentation/power/pm_qos_interface.txt | 6 ++++-- drivers/base/power/qos.c | 22 +++++++++++++++++----- drivers/input/touchscreen/st1232.c | 3 ++- include/linux/pm_qos.h | 7 +++++-- 4 files changed, 28 insertions(+), 10 deletions(-)
include/linux/pm_qos.h +5 -2
@@ -149,7 +149,8 @@ int dev_pm_qos_remove_global_notifier(stvoid dev_pm_qos_constraints_init (struct device *dev);void dev_pm_qos_constraints_destroy (struct device *dev);int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value); +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value);#else static inline enum pm_qos_flags_status __dev_pm_qos_flags (struct device *dev,s32 mask)
@@ -192,7 +193,9 @@ static inline void dev_pm_qos_constraint dev->power.power_state = PMSG_INVALID; }static inline int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type, +s32 value) {return 0 ; }#endif
drivers/base/power/qos.c +17 -5
@@ -565,20 +565,32 @@ EXPORT_SYMBOL_GPL(dev_pm_qos_remove_glob* dev_pm_qos_add_ancestor_request - Add PM QoS request for device's ancestor. * @dev: Device whose ancestor to add the request for. * @req: Pointer to the preallocated handle. +* @type: Type of the request. * @value: Constraint latency value. */ int dev_pm_qos_add_ancestor_request (struct device *dev, -struct dev_pm_qos_request *req,s32 value) +struct dev_pm_qos_request *req, +enum dev_pm_qos_req_type type,s32 value) {struct device *ancestor = dev->parent;int ret = -ENODEV ; -while (ancestor && !ancestor->power.ignore_children) - ancestor = ancestor->parent; +switch (type) { +case DEV_PM_QOS_RESUME_LATENCY : +while (ancestor && !ancestor->power.ignore_children) + ancestor = ancestor->parent; +break ; +case DEV_PM_QOS_LATENCY_TOLERANCE : +while (ancestor && !ancestor->power.set_latency_tolerance) + ancestor = ancestor->parent; + +break ; +default : + ancestor =NULL ; + }if (ancestor) - ret =dev_pm_qos_add_request (ancestor, req, -DEV_PM_QOS_RESUME_LATENCY , value); + ret =dev_pm_qos_add_request (ancestor, req, type, value);if (ret <0 ) req->dev =NULL ;
Documentation/power/pm_qos_interface.txt +4 -2
@@ -134,9 +134,11 @@ The meaning of the return values is as f PM_QOS_FLAGS_UNDEFINED: The device's PM QoS structure has not been initialized or the list of requests is empty. -int dev_pm_qos_add_ancestor_request(dev, handle, value) +int dev_pm_qos_add_ancestor_request(dev, handle, type, value) Add a PM QoS request for the first direct ancestor of the given device whose -power.ignore_children flag is unset. +power.ignore_children flag is unset (for DEV_PM_QOS_RESUME_LATENCY requests) +or whose power.set_latency_tolerance callback pointer is not NULL (for +DEV_PM_QOS_LATENCY_TOLERANCE requests). int dev_pm_qos_expose_latency_limit(device, value) Add a request to the device's PM QoS list of resume latency constraints and
drivers/input/touchscreen/st1232.c +2 -1
@@ -134,7 +134,8 @@ static irqreturn_t st1232_ts_irq_handler }else if (!ts->low_latency_req.dev) {/* First contact, request 100 us latency. */ dev_pm_qos_add_ancestor_request (&ts->client->dev, - &ts->low_latency_req,100 ); + &ts->low_latency_req, +DEV_PM_QOS_RESUME_LATENCY ,100 ); }/* SYN_REPORT */